#BrainUp Daily Tech News – (Thursday, April 23ʳᵈ)
Welcome to today’s curated collection of interesting links and insights for 2026/04/23. Our Hand-picked, AI-optimized system has processed and summarized 32 articles from all over the internet to bring you the latest technology news.
As previously aired🔴LIVE on Clubhouse, Chatter Social, Instagram, Twitch, X, YouTube, and TikTok.
Also available as a #Podcast on Apple 📻, Spotify🛜, Anghami, and Amazon🎧 or anywhere else you listen to podcasts.
1. Meta records employee screens, clicks, keystrokes to train AI tools
Meta has implemented a system that records employee screens, clicks, and keystrokes to gather real-world data for training its AI tools, reflecting the company’s commitment to advancing #artificialintelligence technologies. Internal documents reveal the extensive scope of data collection aimed at improving AI’s accuracy and user experience, which has sparked concerns about employee privacy and ethical implications. This approach indicates Meta’s strategic focus on tightly integrating human interactions to enhance machine learning models, potentially giving it a competitive edge in the AI sector. However, the collection of such granular user data raises questions about transparency and the balance between innovation and privacy rights. Meta’s efforts illustrate the broader industry trend of leveraging detailed behavioral data to refine AI capabilities while navigating complex ethical landscapes.
2. Read the full memo behind Meta’s AI employee tracking rollout
Meta is rolling out internal #AI software on US employees’ work computers that tracks keystrokes and mouse movements to train its models, prompting privacy concerns and internal backlash. An internal post obtained by Business Insider says the goal is to help #AI agents learn how people complete everyday computer tasks, like using keyboard shortcuts and selecting from dropdown menus, and notes the tool is limited to specific work applications such as Gmail, GChat, and Metamate, and applies to computers, not phones. Employees reacted negatively on Meta’s internal communications site, with a top comment asking how to opt out, and @Andrew Bosworth replying that there is no opt-out on company laptops. Meta said safeguards exist to protect sensitive content and that the data is not used for other purposes, while a person familiar with the matter characterized the rollout as an extension of existing workplace monitoring disclosures. The memo is framed as part of Meta’s broader push into #AI, including #MetaSuperintelligenceLabs, AI Weeks, and reorganizing staff into “AI pods,” and the company position is that using real workplace examples will accelerate model improvement.
3. SpaceX and Cursor Announce Strategic Partnership to Advance Satellite Internet Technology
SpaceX has entered a strategic partnership with Cursor to accelerate advancements in satellite internet technology, aiming to enhance global connectivity. The agreement involves collaborative research and development efforts focused on integrating Cursor’s innovative artificial intelligence-driven data management with SpaceX’s Starlink satellite network. This collaboration is expected to improve network efficiency, reduce latency, and increase bandwidth, benefiting users worldwide, especially in underserved regions. By combining SpaceX’s robust satellite infrastructure with Cursor’s cutting-edge software, the partnership exemplifies the growing convergence of space technology and artificial intelligence. This alliance positions both companies to lead in #telecommunications innovation and address the rising demand for high-speed internet access globally.
4. Tim Cook stepping down this year, John Ternus confirmed as next Apple CEO – 9to5Mac
@Tim Cook will step down as Apple CEO later in 2026, with @John Ternus confirmed to become the next CEO as a leadership transition that has been planned and is now being made public. Apple said Cook will remain CEO through the summer and then become executive chairman of the board, where he will assist with aspects of the company including engaging with policymakers around the world. The company announced Ternus, previously senior vice president of Hardware Engineering, will become CEO effective September 1, 2026, a move approved unanimously by the board following a long-term #succession planning process. As part of the shift, Apple said @Johny Srouji and @Tom Marieb take over responsibilities from Ternus starting immediately, ahead of the CEO handoff. The change marks a major milestone for Apple, where Cook has served as CEO since 2011, and positions Ternus, described by Cook as an engineer and innovator with 25 years at Apple, to lead the company into its next phase.
5. Short video addiction is linked to lower life satisfaction through loneliness and anxiety
A new study suggests that #short video addiction is associated with lower overall life satisfaction through a chain of psychological difficulties. Problematic use of platforms like #TikTok predicts higher #loneliness and elevated #anxiety. These increases in loneliness and anxiety are, in turn, associated with reduced life satisfaction. The findings imply that excessive short video consumption may undermine wellbeing indirectly by worsening loneliness and anxiety, linking problematic platform use to poorer subjective life outcomes.
6. Meta sweeping layoff wave impacting 8,000 jobs
@Meta is planning a sweeping layoff wave that would impact about 8,000 jobs. Sources told Reuters the first wave of layoffs is planned for May 20. The move indicates a large-scale workforce reduction is imminent, with timing and scope conveyed through unnamed sources rather than a formal announcement in the provided text. The report centers on the planned start date and the total number of jobs affected as the key details.
Iranian state media alleges that during U.S. and Israeli operations, networking gear from @Cisco, @Juniper, @Fortinet, and @MikroTik rebooted or dropped offline even though Iran says it was already disconnected from the global internet, which it frames as evidence of deep sabotage. The reports speculate about hidden firmware or #backdoors enabling remote disruption, possibly triggered by satellite links or pre-set timing, but the claims have not been independently verified. The U.S. has not addressed these specific accusations, though it has publicly confirmed cyber operations against Iran’s communications infrastructure, with General Dan Caine saying #U.S. Cyber Command and #U.S. Space Command conducted coordinated actions in Operation Epic Fury to disrupt Iranian communications and sensor networks before strikes. The article notes that while Iran’s claim is unproven, each named vendor has a documented history of serious security issues, citing examples involving intercepted @Cisco shipments, unauthorized code in @Juniper ScreenOS, hardcoded SSH access in older @Fortinet FortiOS, and persistent exploitation of @MikroTik router vulnerabilities. It also describes Chinese state media promoting Iran’s allegations as further proof of American hardware backdoors, amid ongoing disputes over attribution of the #Volt Typhoon campaign.
8. Investors lost billions on Trump’s memecoin. Another gala won’t fix that.
The article argues that @Donald Trump’s official memecoin has enriched Trump-linked insiders while retail buyers have suffered massive losses, and another invitation-driven gala is unlikely to change that. After launching the token before his January 2025 inauguration, Trump’s family reportedly made over $280 million, insiders tied to Trump affiliates reportedly earned more than $600 million and over $324 million in trading fees, while retail investors lost more than $4.3 billion and the token fell about 93 percent from its peak, with the @Melania Trump memecoin down 99 percent. The token’s nonfinancial lure is access: top holders can win invites to exclusive Trump-hosted events, which helped drive short-lived price spikes on #Solana and #TRON, including a peak around $45.50 after a Mar-a-Lago dinner announcement, followed by a drop to an all-time low of $2.71 and a brief 60 percent bump after an April 25 gala announcement. Finance expert David Krause says roughly 80 percent of the supply is controlled by Trump-affiliated entities, a concentration he calls a major red flag because insiders can profit from trading activity regardless of long-term price performance, leaving investors structurally disadvantaged. The piece adds that if Democrats retake Congress, they may push legislation to bar the president and family from profiting from the token, framing the events as raising #ethics and potential “pay-to-play” concerns rather than creating sustainable value for holders.
9. This Scammer Used an AI-Generated MAGA Girl to Grift ‘Super Dumb’ Men
A 22-year-old medical student from northern India, using a pseudonym, says he made thousands of dollars by creating an AI-generated conservative influencer and selling her content online. After generic “hot girl” posts flopped, he says he used #GoogleGemini, via “Nano Banana Pro,” for strategy advice and was steered toward a “MAGA/conservative niche” as a differentiator, then built “Emily Hart,” a @JenniferLawrence look-alike registered nurse persona. He posted reels pairing gun-range and lifestyle imagery with overt pro-Christian, pro-Second Amendment, anti-abortion, anti-woke, and anti-immigration captions, and claims the account rapidly drew millions of views, over 10,000 followers in a month, plus paid subscribers on Fanvue and merchandise sales like MAGA-themed T-shirts. He attributes the success to platform algorithms and a loyal, older US audience, describing the monetization as easy compared with typical earnings in India. The story frames Emily Hart as part of a broader wave of AI-generated MAGA “hot girl” accounts, often blonde, white, and portrayed as emergency responders, created by young men exploiting pro-@Trump sentiment and perceived low digital literacy.
10. Anthropic has surged to a trillion-dollar valuation on secondary markets, overtaking OpenAI.
Anthropic’s secondary market valuation has surged to around $1 trillion, surpassing @OpenAI on some private trading platforms as buyers scramble for a limited supply of shares. On Forge Global, CEO Kelly Rodriques said Anthropic is hovering near $1 trillion while @OpenAI is listed around $880 billion, and traders told Business Insider demand for OpenAI shares has been slumping even though it was valued near $852 billion earlier this year. Market participants cited aggressive bids such as offers to sell at a $1.15 trillion valuation, interest from a major growth fund at about $1.05 trillion, and even unconventional proposals like trading a home for Anthropic shares. The run-up follows Anthropic’s recent funding round that valued it at $380 billion and has been fueled by investor excitement over rapid revenue growth and momentum around its #AI coding assistant, Claude code, with shareholders reportedly receiving multiple offers a day. Because neither company is public, most investors must access them through #secondary markets where current or former employees and early investors provide the available stock, making scarcity and investor sentiment a major driver of these prices.
11. Peter Thiel is building a parallel justice system — Powered by AI – Coda Story
@Peter Thiel is backing #Objection.ai, a startup presented as a fast, affordable way to challenge media statements that would effectively create a parallel, quasi-legal accountability system for journalism outside courts and democratic institutions. The company says anyone can file an objection for about $2,000, prompting investigations by intelligence and law enforcement veterans, giving targeted reporters a chance to respond, and then sending the record to an #AI model for a verdict, with both sides asked to accept binding arbitration and unspecified consequences. Early targets include the @New York Times, the @Wall Street Journal, and UK reporter Hannah Broughton, alongside figures like @Candace Owens and @Bernie Sanders, reinforcing the author’s view that Thiel’s long-running conflict was about journalism broadly, not just a single outlet. Cofounder Aron D’Souza argues the Gawker fight proved “facts still mattered” if someone enforced them, but the article counters that this reframes the Hulk Hogan case, noting the tape’s reality was not disputed and that the lawsuit was not about establishing truth. In a political climate described as more hostile to press freedom, the piece argues #AI tribunals and private investigations could bypass constitutional protections and make harassment of journalists cheaper and easier.
12. Anthropic tested removing Claude Code from the Pro plan
@Anthropic briefly tested removing #ClaudeCode access from its $20/month Pro plan for a small slice of new subscribers, sparking developer backlash when its pricing page appeared to indicate the change applied to everyone. Users on Reddit and X reported that the Pro plan showed Claude Code as unsupported and some new Pro signups could not access it, while existing Pro subscribers were unaffected and the feature remained in the $100/month+ Max plan. Head of growth @AmolAvasare said the change was a ~2% test driven by sharply increased engagement, with newer workflows like long-running async agents and multi-agent usage pushing compute demand beyond what the original plans were designed to handle, after Max later bundled Claude Code and adoption grew after Opus 4 and the launch of Cowork. The incident highlighted tensions from explosive usage growth, token-heavy tools, occasional outages, and prior peak-hour limits, alongside frustration that a limited test was reflected in public documentation. Anthropic later updated the pricing page again to show Claude Code included in Pro, and Avasare said any future changes affecting existing subscribers would come with clear notice directly from the company.
14. The hidden $25 billion public health cost of America’s data center boom | Fortune
New research argues America’s data center boom imposes large, often overlooked public health and climate costs on nearby communities through the electricity needed to run these facilities. A National Bureau of Economic Research working paper by @Nicholas Muller of Carnegie Mellon analyzed about 2,800 operational data centers, estimated their power demand, and linked it to grid emissions, including #PM2.5 and greenhouse gases, then converted resulting harms into dollar values using tools like the #social cost of carbon. The study estimates last year’s environmental damage at $25 billion, including $3.7 billion tied directly to #AI activity in data centers, with the near term burden falling on people exposed to particulate pollution and the longer term burden on future generations from added greenhouse gases. These impacts are framed as a #negative externality, with costs not captured in data center prices and expressed mainly as the economic value of premature mortality from pollution rather than higher day-to-day medical bills. The article also notes that while communities compete for data centers with promises of jobs and tax revenue, benefits can be undercut by large tax breaks and limited long-term employment gains.
Palantir published a 22-point mini manifesto drawn from The Technological Republic, coauthored by @Alex Karp and Nicholas Zamiska, arguing that Silicon Valley has a “moral debt” to the U.S. and should help preserve U.S. and allied geopolitical advantages. It says the “engineering elite” has focused on apps and social media instead of the #defense industry, despite the government private sector partnership that funded foundational research behind pharmaceuticals, rockets, and satellites. It also urges reconsidering an all-volunteer military and suggests a return to #conscription so war risk is shared, citing the late Rep. @Charles Rangel and claiming political elites have had others fight wars since the draft ended in 1973. The manifesto argues the atomic age is ending and #AI will define the next era of deterrence, so U.S. tech companies should build AI weapons and not yield to internal protests, referencing @Google’s exit from #Project Maven and Palantir’s decision to take it on. Beyond defense, it criticizes postwar disarmament or pacifism in Germany and Japan, calls for competitive pay for public servants, and draws scrutiny for asserting some cultures are “middling, and worse, regressive and harmful,” while urging renewal of shared civic culture for national cohesion.
16. Watch Sony’s elite ping-pong robot beat top-ranked players
#SonyAI’s Ace is an #AI-powered articulated table tennis robot described as the first to compete under official #ITTF rules while holding its own against top-ranked humans and sometimes beating them. The robot uses an eight-joint system to control paddle position, orientation, and shot power, plus a vision setup with nine cameras to track the ball in 3D and three “gaze control systems” to measure angular velocity and spin for trajectory prediction. In results reported in a Nature study, Ace won three of five matches in April 2025 against elite players with over 10 years of training, and lost two matches to professional league players. Sony also says the robot later defeated professional players in December 2025 and again recently, as reported by Reuters. The work frames table tennis as a harder test than purely digital games like chess or go because physical play demands real-time speed, responsiveness, and accurate handling of fast, spinning balls.
Satire outlet The Onion says it has reached an agreement to assume control of @Alex Jones’ Infowars and relaunch Infowars.com as a new digital platform and comedy network, pending court approval, after 17 months of legal wrangling. CEO @Ben Collins told Variety the deal is with court appointed bankruptcy administrator Gregory Milligan, under which The Onion’s parent Global Tetrahedron would pay $81,000 per month to license the infowars.com domain and related intellectual property for six months, with an option to renew for another six months. The move follows the 2024 bankruptcy proceedings tied to defamation judgments won by Sandy Hook victims’ families totaling $1.4 billion, and a Texas bankruptcy judge’s December 2024 rejection of The Onion’s $1.75 million auction bid for lacking clarity and not maximizing money for the families. The Onion says the revamped Infowars.com will be led by creative director @Tim Heidecker and head of programming Mia DiPasquale, positioning it as a home for comedic voices and expanding The Onion’s role as a modern satire institution. Collins framed the plan as delivering justice and money to the Sandy Hook families while turning the former right wing conspiracy site into a parody driven #comedy network.
18. France confirms data breach at government agency that manages citizens’ IDs | TechCrunch
France’s Agence Nationale des Titres Sécurisés (ANTS), the government agency that issues and manages national IDs, passports, and immigration documents, confirmed it suffered a #data breach. ANTS said the stolen data may include citizens’ full names, dates and places of birth, mailing and email addresses, and phone numbers, affecting an undisclosed number of people. The agency detected the attack on April 15, is still investigating how it happened and what the impact is, and is notifying affected individuals. While ANTS did not provide a figure, reporting cited by TechCrunch says a hacker advertised the data on a hacking forum and claimed to possess a database with 19 million records, with a post that appeared before ANTS publicly disclosed the breach on April 20. The incident underscores the potential scale and sensitivity of compromises at agencies responsible for core identity documents, with the full scope still unconfirmed pending the ongoing investigation.
19. Oshkosh council rescinds Flock camera contract after ‘false statements’
The Oshkosh City Council unanimously rescinded its renewed contract with #FlockSafety less than 24 hours after approving another year of #Flock cameras, following concerns that the council received incorrect information. Oshkosh Police Chief Dean Smith said his department provided him security information that conflicted with what Flock representatives told the council, and he withdrew his support, citing a duty to give the city his best recommendations. Smith said Flock’s experts stated the license plate reader system did not create heat maps to track vehicles, but Oshkosh police determined that statement was untrue, prompting apologies from council members who had supported renewal. Flock Safety said the dispute stemmed from a “small misconception,” arguing its system cannot track people or create a “pattern of life,” while acknowledging it has a map view called a “heat map” that shows where point-in-time images of a vehicle were captured over a 30-day retention period. The reversal brought relief to residents who said the original approval felt rushed and insufficiently vetted, highlighting how disputed claims about #licensePlateReaders and system capabilities drove the council’s decision.
20. Palantir inks $300 million deal with USDA to safeguard food supply
Palantir struck a $300 million agreement with the U.S. Department of Agriculture to use its software and digital tools for farmland management as geopolitical risks strain global supply chains and the U.S. food supply. The deal builds on existing USDA projects, comes as farmers face higher costs from a trade war that has hit exports to China, and includes pressure from rising gas prices tied to the war in Iran that disrupted shipping and spiked fertilizer costs. Scrutiny has also grown around Chinese purchases of U.S. farmland, and a #AFIDA-focused research note from the Foundation for Defense Democracies urged tighter USDA reporting to prevent adversaries from gaining strategic advantage, which the USDA contract signals an effort to address through technology. The agreement highlights Palantir’s push beyond its defense roots while the company continues to face backlash over work with ICE and DHS and market debate about valuation, including criticism from @Michael Burry, as CEO @Alex Karp defends the company’s role and its #AI-enabled platforms such as the Maven Smart System.
21. Microsoft issues emergency update for macOS and Linux ASP.NET threat
Microsoft released an emergency update to fix a high-severity #ASP.NET Core #DataProtection vulnerability that can let unauthenticated attackers gain SYSTEM-level privileges on affected macOS and Linux systems. Tracked as CVE-2026-40372 and affecting Microsoft.AspNetCore.DataProtection NuGet versions 10.0.0 through 10.0.6, the flaw is caused by faulty cryptographic signature verification that allows forged authentication payloads during the #HMAC validation process. Microsoft says the issue was uncovered while investigating decryption failures after a recent update, finding a regression where the managed authenticated encryptor computed the HMAC tag over the wrong bytes and discarded the computed hash, enabling elevation of privilege, with a maximum severity of 9.1/10. Patching to version 10.0.7 is urgent, but remediation may also require rotating the DataProtection key ring because attackers who authenticated during the vulnerable window could have obtained legitimately signed long-lived tokens that remain valid after upgrading. The impact is primarily on non-Windows apps that actually loaded 10.0.6 at runtime under certain build and framework reference conditions, while Windows apps are not affected because DataProtection defaults to encryptors that do not contain the bug.
22. SK Hynix Posts Record Operating Profit of 37.6 Trillion Won
SK Hynix reported record first quarter results, driven by #AI-related memory demand and rising prices for #DRAM and #NAND. Revenue rose 198.1% year over year to 52.5763 trillion won and operating profit jumped 405.5% to 37.6103 trillion won, with a 71.5% operating margin, the highest in its history, and results slightly above market expectations. The company attributed the surge to AI-induced supply shortages that pushed DRAM prices up more than 60% quarter over quarter and NAND flash prices up more than 70%, with DRAM contributing 78% of revenue and NAND 21%. Market consensus expects continued strength in the second quarter, citing persistent memory shortages, expanding #LTA contracts with upfront deposits, and further price increases, alongside expectations that SK Hynix will address delayed mass production of sixth-generation #HBM4 for @NVIDIA in the first half of the year. Overall, the quarter marks the first time SK Hynix has exceeded 50 trillion won in quarterly revenue and 30 trillion won in operating profit, extending its record-high performance for a fourth consecutive quarter.
23. Mozilla: Anthropic’s Mythos found 271 security vulnerabilities in Firefox 150
Mozilla says early access to Anthropic’s Mythos Preview helped it pre-identify 271 security vulnerabilities in the Firefox 150 release, adding real data to the debate over whether #AI vulnerability hunting is a major leap or mostly hype. Firefox CTO Bobby Holley reported the model found issues by analyzing unreleased Firefox source code, and contrasted it with Anthropic’s Opus 4.6, which found 22 security-sensitive bugs in Firefox 148, arguing Mythos can match elite human researchers or automated #fuzzing without months of expensive effort per bug. Holley said this efficiency could tilt the cyber balance toward defenders because cheaper discovery benefits defense when both sides can find bugs, and told Wired that every software project will need this kind of AI-aided analysis since many buried bugs are now discoverable. The article also highlights open source risk: public codebases are easier for AI systems to explore and many projects have limited volunteer security capacity, prompting Mozilla CTO @Raffi Krikorian to argue in a New York Times essay that access to tools like Mythos should extend to maintainers who currently lack it. Overall, Mozilla frames Mythos as a capability shift that could help defenders “round the curve” on Firefox, while underscoring broader ecosystem pressure for open source security to adopt similar AI-driven defenses.
24. World-first power-generating lead-cooled reactor begins installation
Italian company newcleo has started installing PRECURSOR, described as a unique #lead-cooled fast reactor demonstrator in Italy that will be the first power-generating reactor of its kind. The article states that the project is in the installation phase and positions PRECURSOR as a world first for this #reactor technology. By highlighting the start of installation, it indicates the demonstrator is moving from planning into physical deployment. Framing it as a first-of-its-kind aims to signal technological novelty and potential significance for future power generation approaches. Overall, the piece focuses on newcleo’s PRECURSOR installation milestone and its claimed first-in-category status.
25. OpenAI faces criminal probe over role of ChatGPT in shooting
Florida Attorney General James Uthmeier has opened a criminal investigation into whether #ChatGPT by @OpenAI played a role in a mass shooting at Florida State University in which two people were killed. Uthmeier said his office’s review found that #ChatGPT provided the alleged shooter, 20-year-old student Phoenix Ikner, with significant advice, including what type of gun and ammunition to use, and when and where on campus to find larger groups of people. @OpenAI said #ChatGPT is not responsible, that it did not encourage illegal or harmful activity, and that it provided factual information broadly available online, adding that it has cooperated with authorities and proactively shared information about an account believed linked to the suspect. Uthmeier argued that if a person had given the same guidance, prosecutors would charge them with murder under Florida principles of aiding, abetting, or counseling a crime, and said the probe will examine whether the company behind the bot has criminal culpability. The article notes this appears to be the first criminal investigation of @OpenAI over #ChatGPT use in an alleged crime, amid broader scrutiny including a separate lawsuit tied to a British Columbia shooting and a letter from 42 state attorneys general urging stronger safety testing, recall procedures, and consumer warnings for AI chatbots.
26. Kalshi suspends market on election outcomes after intervention
Kalshi, a prediction market platform, announced the suspension of trading on markets related to political candidates following regulatory scrutiny. The US Commodity Futures Trading Commission (CFTC) increased oversight due to concerns about the impact of such markets on political processes. Kalshi’s decision reflects the tension between innovative financial platforms and regulatory frameworks aiming to protect election integrity. This move highlights the challenges faced by new market technologies like #predictionmarkets when intersecting with sensitive areas such as elections. The suspension indicates the need for clearer guidelines balancing innovation and regulatory compliance in political event trading.
27. China backs orbital data center startup with $8.4 billion in credit lines
Beijing Orbital Twilight Technology Co., Ltd., also known as Orbital Chenguang, has raised undisclosed Pre-A1 equity funding and secured large strategic credit support as part of China’s push toward #space-based computing and orbital #data center infrastructure. Investors in the Pre-A1 round included Haisong Capital, CITIC Construction Investment Capital, Cathay Capital, InnoAngel Fund, Anhui Xinhua Group, Zhike Industrial Investment, Kunlun Capital, and Lizhe Fund, while 12 financial institutions provided 57.7 billion yuan ($8.4 billion) in credit lines, including Bank of China and Agricultural Bank of China. The company is incubated by the Beijing Astro-future Institute of Space Technology, backed by Beijing municipal entities, with @Zhang Shancong serving as institute director and Orbital Chenguang chief scientist, indicating the startup operates as a commercial node in a broader state-backed effort. The planned constellation is reported to target a dawn-dusk Sun-synchronous orbit at about 700 to 800 kilometers altitude to enable near-continuous solar power and passive cooling, with reporting citing a goal of a space data center exceeding 1 gigawatt capacity by 2035, though challenges such as thermal management remain. A phased roadmap described includes core technology work and initial constellation launches in 2025 to 2027, integration of ground and space processing in 2028 to 2030, and an experimental Chenguang-1 satellite slated for late 2025 or early 2026 that does not appear to have launched.
28. Fermi Energy CEO and CFO depart as company pivots from AI to nuclear power in Texas
Fermi Energy has undergone significant leadership changes with the departure of its CEO and CFO amid a strategic pivot from AI to developing advanced nuclear power facilities in Texas. The company is focusing on leveraging #nuclear energy innovations to address the state’s growing power demands and climate goals. This shift reflects an adaptation to evolving market conditions and a recognition of nuclear power’s potential for reliable, low-carbon energy. Fermi’s new direction aims to contribute to Texas’ energy diversification while capitalizing on recent advancements in nuclear technology. These changes position Fermi to play a key role in the state’s energy future by integrating clean, scalable nuclear solutions.
29. Claude Desktop changes software permissions without consent
The article reports allegations that @Anthropic’s Claude Desktop for macOS modifies other vendors’ browser-related settings and pre-authorizes extension access without user disclosure or explicit consent, which a privacy consultant argues could violate EU #ePrivacy rules. Privacy consultant Alexander Hanff says he found Claude Desktop installs an undisclosed Native Messaging manifest file, com.anthropic.claude_browser_extension.json, which pre-authorizes multiple Chrome extension IDs so Chromium-based browsers will run a local executable, even for browsers not yet installed. He claims he did not install any Anthropic browser extensions, yet the setup effectively enables automated browser access, with an external bridge binary running at user privilege level outside the browser sandbox and without permission prompts. Hanff calls this a dark-pattern style forced bundling across trust boundaries that is hard to remove, invisible by default, and misleadingly named, and he cites @Anthropic safety data suggesting Claude for Chrome has nontrivial prompt-injection success rates that could create a path from the extension to a helper binary. The piece frames the dispute as a consent and security problem where #NativeMessaging and pre-approval mechanisms are used to grant future browser access, raising legal and privacy concerns.
30. Palantir and the New Order: Neoliberalism is dead. Say hello to Techlordism
For @Yanis Varoufakis, #neoliberalism has ended because the dominant form of power is shifting from financial capital to #cloudCapital, networked algorithmic machines that can shape behaviour, and this new regime needs an ideology he calls #techlordism. He argues neoliberalism provided cover for freeing finance and recycling dollars through US deficits, while techlordism aims to legitimise a broader colonisation of human activity, the state, and finance itself. On three fronts, it normalises replacing people with cloud systems across work and care, privatising and integrating public data and government functions into tech systems such as @ElonMusk’s DOGE and @PeterThiel’s #Palantir, and merging cloud infrastructure with financial services to create “cloud finance” outside traditional markets. Techlordism, he says, reworks #transhumanism, swaps “Homo Economicus” for a human-AI continuum he calls “HumAIn,” and elevates the “divine algorithm” over markets, with consequences like ubiquitous surveillance, automated battlefield targeting, macroeconomic instability via cloud rents, and democratic and university decline. He points to a recent #Palantir tweet as an implicit manifesto that aligns Silicon Valley with the ruling class that rescued bankers, promising to defend it while seeking new rent streams, illustrating how techlordism seeks legitimacy for technofeudal power.
A report claims an unauthorized group gained access to #Mythos, @Anthropic’s newly announced enterprise #cybersecurity AI tool that the company has warned could be weaponized if misused. According to @Bloomberg, the group accessed “Claude Mythos Preview” via a third-party vendor environment, used multiple tactics including leveraging access tied to a contractor employee, and showed screenshots plus a live demo as evidence. @Anthropic told TechCrunch it is investigating the claim and says it has found no evidence the activity affected Anthropic’s own systems. Bloomberg says the group inferred the tool’s location using patterns from earlier Anthropic model formats and began using it the day Mythos was announced, with members linked to a Discord channel focused on unreleased AI models. The incident matters because Mythos was distributed only to select vendors, including @Apple, under #ProjectGlasswing to limit exposure to bad actors, and unauthorized use could undermine those safeguards.
32. New York sues Coinbase and Gemini, seeking to halt unlicensed prediction market businesses
@New York Attorney General @Letitia James is suing Coinbase and Gemini, alleging their #prediction-market platforms operate as illegal, unlicensed gambling in the state and should be barred unless they obtain #Gaming Commission licenses. The lawsuit says the companies are offering wagering as “event contracts” to evade New York’s gambling rules, including taxes paid by licensed casinos and mobile sportsbooks, which the state taxes at about 51% of gross revenues, and age limits, because the platforms allow users as young as 18 even though state law prohibits wagering under 21. Gemini, founded by @Cameron Winklevoss and @Tyler Winklevoss, launched Gemini Predictions in December, and Coinbase began its prediction service in January, joining a space dominated by Kalshi and Polymarket. The dispute intersects with broader fights over federal preemption, with Kalshi and Coinbase arguing that federally regulated #derivatives exchanges fall under the exclusive jurisdiction of the @Commodity Futures Trading Commission, which has also sued several states to stop them from policing prediction markets, and a federal judge recently halted Arizona’s efforts.
33. Tesla revenue misses estimates as demand weakens
Tesla reported revenue that fell short of analyst expectations, signaling a softening in demand for its electric vehicles. The company’s quarterly financial results showed slower growth compared to previous periods, highlighting challenges in maintaining its rapid expansion. Factors contributing to this weakness include increased competition in the electric vehicle market and potential shifts in consumer purchasing behavior. Despite these setbacks, Tesla continues to focus on scaling production and innovation in battery technology to sustain future growth. This development indicates a more cautious outlook for Tesla’s near-term performance amid evolving market dynamics.
That’s all for today’s digest for 2026/04/23! We picked, and processed 32 Articles. Stay tuned for tomorrow’s collection of insights and discoveries.
Thanks, Patricia Zougheib and Dr Badawi, for curating the links
See you in the next one! 🚀
