Today's Top AI and Tech News
Top Stories
US government asks OpenAI to stagger GPT-5.6 release with approved partners first
AI helps researchers read a 2,000-year-old Vesuvius scroll charred by Mount Vesuvius eruption
DuckDuckGo AI falsely claims Trump died of rabies, exposing search tool vulnerabilities
IBM unveils world's first sub-1nm chip with nanostack architecture, doubling transistor density
AI Security
THE OUTPOST INSIGHTThe US government is now deciding who gets access to AI models before they launch, formalizing a tiered system where approval determines market entry.
US government asks OpenAI to stagger GPT-5.6 release with approved partners first
The Trump administration has asked OpenAI to limit the launch of its next model, GPT-5.6, releasing it first to government-approved partners before broader availability. This marks the first time the US government has preemptively intervened to restrict a model's launch before it happens, signaling a shift toward active federal oversight of frontier AI releases.
Trump Administration Presses OpenAI to Limit GPT-5.6 Release to Select Partners
The Trump administration has asked OpenAI to restrict GPT-5.6 access to government-approved partners during its initial rollout, citing potential security risks. CEO Sam Altman told employees the government will approve access customer by customer, with a broader public release planned weeks later if the preview period goes smoothly.
India AI ambitions face setback as US export controls suspend access to Anthropic AI models
A US government directive abruptly suspended India's access to Anthropic's advanced Mythos and Fable models, exposing critical vulnerabilities in the country's AI strategy. The incident has sparked urgent discussions between Washington and New Delhi over stable AI access, with experts warning that reliance on foreign-controlled AI threatens India's goal of becoming an AI power rather than merely an adopter.
North Korean Hackers Deploy Gaslight Malware to Deceive AI Security Tools with Fake Errors
A newly discovered macOS malware called Gaslight uses prompt injection to confuse AI-assisted analysis platforms. Attributed to North Korea-aligned threat actors, the Rust-based implant embeds 38 fabricated system messages designed to make AI triage agents abort or refuse analysis. The malware also functions as an information stealer with Telegram-based command-and-control capabilities.
Technology
THE OUTPOST INSIGHTOpenAI's Jalapeño chip arrives as Amazon and Google prepare to sell their own silicon, turning big AI customers into Nvidia's direct competitors.
OpenAI and Broadcom unveil Jalapeño chip to power next-generation AI inference at scale
Google launches Study Notebooks in Gemini AI to transform how students learn and prepare for exams
AI Agents Turn to Simulated Worlds as Large Language Models Hit Practical Limits
OpenAI unveils Jalapeño chip to reduce dependence on Nvidia and make ChatGPT cheaper to run
Anthropic
THE OUTPOST INSIGHTIndia is racing to build sovereign AI infrastructure while 96% of government leaders cite geopolitical risk, then US export controls prove exactly why.
India AI ambitions face setback as US export controls suspend access to Anthropic AI models
A US government directive abruptly suspended India's access to Anthropic's advanced Mythos and Fable models, exposing critical vulnerabilities in the country's AI strategy. The incident has sparked urgent discussions between Washington and New Delhi over stable AI access, with experts warning that reliance on foreign-controlled AI threatens India's goal of becoming an AI power rather than merely an adopter.
EU negotiates with White House after US export ban cuts off Anthropic's most advanced AI models
Brussels held talks with Washington after a US export-control order abruptly cut Europe off from Anthropic's Fable 5 and Mythos 5 models in mid-June. The ban affected European governments, companies, and research institutions, including ENISA and NATO, forcing the EU to negotiate access through the White House rather than directly with Anthropic.
Anthropic accuses Alibaba of massive attack to clone Claude using 28.8 million exchanges
Anthropic has accused Alibaba of conducting the largest known distillation attack on its Claude AI model, generating over 28.8 million exchanges through nearly 25,000 fraudulent accounts between April and June. The company warns this represents a national security threat and is urging Congress to pass legislation penalizing such behavior to protect US AI leadership.
Anthropic accuses Alibaba of using 25,000 fake accounts to steal Claude AI capabilities
Anthropic has accused Alibaba of orchestrating the largest known attack against its systems, allegedly using nearly 25,000 fraudulent accounts to extract capabilities from Claude AI models. The campaign generated 28.8 million queries between April and June, employing a technique called distillation to potentially replicate advanced AI capabilities at lower cost.
Sovereign AI
THE OUTPOST INSIGHTThe EU must now negotiate AI access with the White House instead of buying from companies, turning American AI into a diplomatic bargaining chip.
EU negotiates with White House after US export ban cuts off Anthropic's most advanced AI models
Brussels held talks with Washington after a US export-control order abruptly cut Europe off from Anthropic's Fable 5 and Mythos 5 models in mid-June. The ban affected European governments, companies, and research institutions, including ENISA and NATO, forcing the EU to negotiate access through the White House rather than directly with Anthropic.
India AI ambitions face setback as US export controls suspend access to Anthropic AI models
A US government directive abruptly suspended India's access to Anthropic's advanced Mythos and Fable models, exposing critical vulnerabilities in the country's AI strategy. The incident has sparked urgent discussions between Washington and New Delhi over stable AI access, with experts warning that reliance on foreign-controlled AI threatens India's goal of becoming an AI power rather than merely an adopter.
China AI shifts focus to affordability over capability in race against US frontier models
Chinese AI companies are redefining the US China AI race by prioritizing widespread adoption through affordable models rather than technical superiority. Firms like Zhipu AI and Alibaba are gaining contracts across Asia and the Middle East, offering 80% of leading US capabilities at significantly lower costs while US export restrictions create market openings.
TensorX and Solstice announce €1bn partnership to build sovereign AI infrastructure across Europe
Dublin-based TensorX and finance provider Solstice have partnered to deliver up to €1bn in sovereign European AI infrastructure. The collaboration will finance AI hardware and data-centre build-outs to meet rising EU sovereign AI demand. Solstice will launch aiUSX, a yield-bearing asset enabling companies to finance AI infrastructure with capital they already hold.
Business and Economy
THE OUTPOST INSIGHTMicron's 81% margin reveals AI companies are so desperate for memory chips they'll pay almost any price, reshaping the entire semiconductor power structure.
Micron revenue quadruples to $41.46 billion as AI boom drives memory chip shortage through 2027
Apple hikes Mac and iPad prices up to 25% as AI boom sends memory chip costs soaring
Apple raises Mac and iPad prices up to $500 as AI memory shortage hits consumer electronics
The tokenmaxxing era ends as companies struggle to measure AI effectiveness amid spiraling costs
Claude
THE OUTPOST INSIGHTAnthropic warns Congress about Alibaba cloning Claude while the White House restricts Claude's export to European allies, including NATO.
Anthropic accuses Alibaba of massive attack to clone Claude using 28.8 million exchanges
Anthropic has accused Alibaba of conducting the largest known distillation attack on its Claude AI model, generating over 28.8 million exchanges through nearly 25,000 fraudulent accounts between April and June. The company warns this represents a national security threat and is urging Congress to pass legislation penalizing such behavior to protect US AI leadership.
Anthropic accuses Alibaba of using 25,000 fake accounts to steal Claude AI capabilities
Anthropic has accused Alibaba of orchestrating the largest known attack against its systems, allegedly using nearly 25,000 fraudulent accounts to extract capabilities from Claude AI models. The campaign generated 28.8 million queries between April and June, employing a technique called distillation to potentially replicate advanced AI capabilities at lower cost.
Anthropic's White House Negotiations Improve After CEO Dario Amodei Replaced by Co-Founder
Anthropic CEO Dario Amodei has been sidelined from critical White House negotiations over export controls on the company's advanced AI models. Co-founder Tom Brown and public policy chief Sarah Heck are now leading talks with the Trump Administration, with sources indicating discussions are progressing better since the leadership change.
Anthropic accuses Alibaba of largest distillation attack to extract Claude AI model capabilities
U.S. AI company Anthropic has accused Chinese tech giant Alibaba of conducting the largest known distillation attack to illicitly extract Claude AI model capabilities. The campaign involved 28.8 million exchanges through nearly 25,000 fraudulent accounts between April and June 2026, targeting commercially valuable features like software engineering and agentic reasoning.
Startups
THE OUTPOST INSIGHTQualcomm is spending up to $14 billion on two AI chip startups while its CEO predicts apps will disappear entirely.
Qualcomm acquires AI chip startup Modular for $4 billion to challenge Nvidia's dominance
Runpod raises $100M at $1 billion valuation as AI developers scramble for computing power
Sail Research raises $80M to slash costs for running AI agents by up to 10 times
Scaled Cognition raises $100M from Khosla to eliminate AI hallucinations with new architecture
IBM
THE OUTPOST INSIGHTIBM's vertical transistor stacking and SanDisk's proposal to stack memory beneath processors suggest the chip industry is building upward because it can't shrink sideways anymore.
IBM unveils world's first sub-1nm chip with nanostack architecture, doubling transistor density
IBM has developed the world's first sub-1nm computer chip using a revolutionary nanostack architecture that vertically stacks transistors. The fingernail-sized chip packs nearly 100 billion transistors—double the density of current 2nm chips—while delivering 50% better performance and 70% improved energy efficiency. This breakthrough could reshape AI computing and semiconductor manufacturing over the next decade.
IBM chip breaks 1nm barrier with nanostack architecture, packing 100 billion transistors
IBM unveiled the first sub-1 nanometer chip using its nanostack architecture, fitting nearly 100 billion transistors into a fingernail-sized die. The 0.7nm chip delivers up to 50% better performance and 70% greater energy efficiency compared to IBM's 2nm version. Production is expected within five years as demand for energy-efficient hardware accelerates.
IBM unveils sub-nanometer chip with 100 billion transistors to power next-gen AI computing
IBM introduced the world's first sub-1 nanometer chip technology using a nanostack architecture that packs nearly 100 billion transistors on a fingernail-sized chip. The 0.7-nanometer node design promises 50% higher performance or 70% greater energy efficiency compared to previous generations, addressing growing demands for AI workloads in data centers.
IBM and OpenAI partner to deploy frontier AI models in enterprise security operations
IBM has joined OpenAI's Daybreak Cyber Partner Program to integrate frontier AI capabilities into corporate cybersecurity workflows. The partnership launches with a new application security service that uses OpenAI's models to detect and validate software vulnerabilities faster than traditional tools, backed by a $5 billion commitment through Project Lightwell to secure open-source software.
Copilot
THE OUTPOST INSIGHTMicrosoft's Satya Nadella warns AI can't hollow out industries while the New York Times and 400 newspapers sue him for doing exactly that.
OpenAI and Microsoft Face Expanded Copyright Lawsuit as 400 Newspapers Join Legal Battle
The New York Times amended its copyright lawsuit against OpenAI and Microsoft, accusing Microsoft of actively encouraging the use of copyrighted articles for AI training. Meanwhile, 400 newspapers filed a separate lawsuit alleging content theft to build ChatGPT and Microsoft Copilot. The legal battles highlight mounting pressure on AI companies over their use of protected content without permission or compensation.
Italy antitrust probes Microsoft over alleged AI bundling and Microsoft 365 price hike practices
Italy's competition authority has opened a Microsoft investigation into claims the tech giant quietly bundled AI features into Microsoft 365 and defaulted subscribers onto pricier plans. The Italian competition watchdog alleges Microsoft failed to clearly communicate changes involving Copilot and Designer, potentially violating consumer rules and restricting choice through aggressive pricing strategies.
Italy launches antitrust probe into Microsoft's AI bundling and automatic price increases
Italy's competition authority has opened an investigation into Microsoft over alleged unfair commercial practices tied to its Microsoft 365 subscription service. The regulator claims customers were automatically moved to a pricier Copilot-bundled plan unless they actively opted out, with insufficient information provided to make informed decisions about the AI price rise taking effect from July 2026.
Microsoft Copilot in Excel gains reusable skills and live data connectors for finance teams
Microsoft rolled out major updates to Copilot in Excel aimed at finance professionals, introducing reusable skills for automating repeatable workflows like DCF modeling and variance analysis. The update includes new live data connectors from providers like FactSet, S&P Global, PitchBook, and Morningstar, plus enhanced traceability features that track every AI-generated change in spreadsheets.
Office 365
THE OUTPOST INSIGHTMicrosoft is automating the DCF models and variance analysis that justify finance jobs, not just making spreadsheets easier.
Microsoft Copilot in Excel gains reusable skills and live data connectors for finance teams
Microsoft rolled out major updates to Copilot in Excel aimed at finance professionals, introducing reusable skills for automating repeatable workflows like DCF modeling and variance analysis. The update includes new live data connectors from providers like FactSet, S&P Global, PitchBook, and Morningstar, plus enhanced traceability features that track every AI-generated change in spreadsheets.
Italy antitrust probes Microsoft over alleged AI bundling and Microsoft 365 price hike practices
Italy's competition authority has opened a Microsoft investigation into claims the tech giant quietly bundled AI features into Microsoft 365 and defaulted subscribers onto pricier plans. The Italian competition watchdog alleges Microsoft failed to clearly communicate changes involving Copilot and Designer, potentially violating consumer rules and restricting choice through aggressive pricing strategies.
Italy launches antitrust probe into Microsoft's AI bundling and automatic price increases
Italy's competition authority has opened an investigation into Microsoft over alleged unfair commercial practices tied to its Microsoft 365 subscription service. The regulator claims customers were automatically moved to a pricier Copilot-bundled plan unless they actively opted out, with insufficient information provided to make informed decisions about the AI price rise taking effect from July 2026.
Microsoft patches critical Copilot flaw that let hackers steal 2FA codes with one click
Microsoft fixed a max-severity vulnerability in its M365 Copilot AI platform that allowed attackers to exfiltrate 2FA codes and sensitive data through a crafted URL. Security researchers at Varonis discovered the exploit chain, dubbed SearchLeak, which combined prompt injection with HTML rendering flaws to bypass Microsoft's guardrails. The vulnerability highlights how AI systems struggle to distinguish between legitimate user instructions and malicious commands embedded in third-party content.
Responsible AI
THE OUTPOST INSIGHTPrivacy-focused DuckDuckGo and dominant Google are both falling for the same AI search flaw: treating coordinated fiction as verified fact.
DuckDuckGo AI falsely claims Trump died of rabies, exposing search tool vulnerabilities
DuckDuckGo's AI-powered search tool mistakenly reported that President Donald Trump died from rabies, falling victim to a coordinated misinformation campaign by Reddit users. The incident highlights critical vulnerabilities in AI-generated content and raises questions about the reliability of AI-powered search results from privacy-focused platforms.
Trump Administration Presses OpenAI to Limit GPT-5.6 Release to Select Partners
The Trump administration has asked OpenAI to restrict GPT-5.6 access to government-approved partners during its initial rollout, citing potential security risks. CEO Sam Altman told employees the government will approve access customer by customer, with a broader public release planned weeks later if the preview period goes smoothly.
Tim Sweeney blasts Valve's AI tags on Steam, calls them irresponsible as games face backlash
Epic Games CEO Tim Sweeney criticized Valve for requiring AI disclosure tags on Steam, arguing developers shouldn't be penalized for using AI tools. Games with AI tags receive 53% fewer reviews, creating what Sweeney calls a 'Scarlet Letter' that makes success harder for developers using productivity-enhancing AI.
Hasbro Faces Backlash Over AI Voice Rights Demands for Child Actors in Peppa Pig Contracts
Hasbro is under fire after reportedly requiring child voice actors on Peppa Pig to sign away their voices for AI use. The Agents of Young Performers Association released an open letter with over 1,000 signatures arguing children cannot provide informed consent for such agreements. The controversy highlights growing tensions over generative AI use in the entertainment industry.
THE OUTPOST INSIGHTGoogle builds AI that diagnoses student knowledge gaps while educators struggle with AI detection tools that fail 99.6% of the time.
Google launches Study Notebooks in Gemini AI to transform how students learn and prepare for exams
Google has introduced Study Notebooks in its Gemini AI app, an AI-powered feature that creates personalized learning experiences for students. Instead of simply answering questions, the tool identifies knowledge gaps through diagnostic quizzes and builds customized lessons that adapt in real-time. The feature includes standardized test prep for exams like SAT, JEE, NEET, and GRE, with integration into NotebookLM for additional study aids.
Google Finance exits beta with dedicated Android app and AI-powered portfolio tracking tools
Google launched a dedicated Android app for Google Finance, marking the platform's exit from beta after AI-powered redesign. The app provides real-time market data, live financial news, and AI research tools with Key Moments explaining stock movements. New portfolio tracking features let users consolidate investments in one dashboard, while scheduled briefings deliver custom market analysis. An iOS version is coming later this year.
Google Finance launches first Android app with AI tools after 20 years as web-only platform
After two decades as a web-only service, Google Finance has released its first standalone mobile app exclusively for Android, with iOS coming later this year. The launch coincides with the AI-powered overhaul of the Finance website leaving beta, bringing generative AI features like conversational stock analysis, Key Moments explanations for market fluctuations, and portfolio insights to mobile users.
Google integrates computer use into Gemini 3.5 Flash, enabling AI agents to control screens
Google has integrated computer use capability directly into Gemini 3.5 Flash, allowing AI agents to see screens, navigate browsers, and perform tasks autonomously. Previously requiring a standalone Gemini 2.5 model, the feature now works through the Gemini API and Enterprise Agent Platform with new enterprise safeguards against prompt injection attacks.
Videos
OpenAI Unveils First Custom AI Chip With Broadcom | Bloomberg Tech 6/24/2026
Midjourney wants to delete 30% of all death...
AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
Stay ahead of the curve. Get the latest AI news, delivered to your inbox.
Subscribe to our newsletter
Get the latest updates delivered to your inbox every day, and stay up-to-date for free 🧠📈
Did you know?
Constitutional AI
Constitutional AI is a training approach developed by Anthropic where AI systems learn to follow a set of predefined rules rather than relying solely on human feedback for every decision. The AI critiques and revises its own outputs based on this constitution.
Videos
OpenAI Unveils First Custom AI Chip With Broadcom | Bloomberg Tech 6/24/2026
Bloomberg TechnologyMidjourney wants to delete 30% of all death...
FireshipAI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
Latent SpaceFrom Your Feed / Also in the News
OpenAI and Broadcom unveil Jalapeño chip to power next-generation AI inference at scale
OpenAI and Broadcom have announced Jalapeño, a custom AI processor designed specifically for large language model inference in data centers. The chip promises better performance per watt than current options and marks OpenAI's push toward vertical integration. With deployment planned by year-end, this move signals OpenAI's strategy to reduce dependence on Nvidia while controlling its entire AI infrastructure stack.
THE OUTPOST INSIGHTOpenAI's Jalapeño chip arrives as Amazon and Google prepare to sell their own silicon, turning big AI customers into Nvidia's direct competitors.
Google launches Study Notebooks in Gemini AI to transform how students learn and prepare for exams
China AI shifts focus to affordability over capability in race against US frontier models
Chinese AI companies are redefining the US China AI race by prioritizing widespread adoption through affordable models rather than technical superiority. Firms like Zhipu AI and Alibaba are gaining contracts across Asia and the Middle East, offering 80% of leading US capabilities at significantly lower costs while US export restrictions create market openings.
THE OUTPOST INSIGHTChinese AI firms are winning markets by doing exactly what Microsoft's Nadella says the industry needs: cheaper models and wider access, while US companies restrict releases and chase technical supremacy.
TensorX and Solstice announce €1bn partnership to build sovereign AI infrastructure across Europe
Did you know?
Constitutional AI
Constitutional AI is a training approach developed by Anthropic where AI systems learn to follow a set of predefined rules rather than relying solely on human feedback for every decision. The AI critiques and revises its own outputs based on this constitution.
Anthropic's White House Negotiations Improve After CEO Dario Amodei Replaced by Co-Founder
Anthropic CEO Dario Amodei has been sidelined from critical White House negotiations over export controls on the company's advanced AI models. Co-founder Tom Brown and public policy chief Sarah Heck are now leading talks with the Trump Administration, with sources indicating discussions are progressing better since the leadership change.
THE OUTPOST INSIGHTAnthropic's negotiations improved only after swapping its CEO for a co-founder, suggesting the White House is now vetting not just AI models but the executives who pitch them.
Anthropic accuses Alibaba of largest distillation attack to extract Claude AI model capabilities
IBM unveils world's first sub-1nm chip with nanostack architecture, doubling transistor density
IBM has developed the world's first sub-1nm computer chip using a revolutionary nanostack architecture that vertically stacks transistors. The fingernail-sized chip packs nearly 100 billion transistors—double the density of current 2nm chips—while delivering 50% better performance and 70% improved energy efficiency. This breakthrough could reshape AI computing and semiconductor manufacturing over the next decade.
THE OUTPOST INSIGHTIBM's vertical transistor stacking and SanDisk's proposal to stack memory beneath processors suggest the chip industry is building upward because it can't shrink sideways anymore.
IBM chip breaks 1nm barrier with nanostack architecture, packing 100 billion transistors
Microsoft patches critical Copilot flaw that let hackers steal 2FA codes with one click
Microsoft fixed a max-severity vulnerability in its M365 Copilot AI platform that allowed attackers to exfiltrate 2FA codes and sensitive data through a crafted URL. Security researchers at Varonis discovered the exploit chain, dubbed SearchLeak, which combined prompt injection with HTML rendering flaws to bypass Microsoft's guardrails. The vulnerability highlights how AI systems struggle to distinguish between legitimate user instructions and malicious commands embedded in third-party content.
THE OUTPOST INSIGHTMicrosoft patched SearchLeak while researchers expose OpenClaw and Agentjacking flaws, revealing AI tools are now both the security threat and the attack surface.
Google Finance exits beta with dedicated Android app and AI-powered portfolio tracking tools
Google launched a dedicated Android app for Google Finance, marking the platform's exit from beta after AI-powered redesign. The app provides real-time market data, live financial news, and AI research tools with Key Moments explaining stock movements. New portfolio tracking features let users consolidate investments in one dashboard, while scheduled briefings deliver custom market analysis. An iOS version is coming later this year.
THE OUTPOST INSIGHTGoogle Finance launches an Android app years after killing its standalone app in 2017, betting AI features can win back users who moved to competitors.
Google Finance launches first Android app with AI tools after 20 years as web-only platform
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
