OpenAI expands Codex with role-specific plugins for data analysis, sales, and investment banking. 5 million weekly users, 20% non-developers. This segment grows 3x faster than developers.
Anthropic scales Project Glasswing to 150 partners across 15+ countries using Claude Mythos Preview to detect critical flaws. Existing partners have identified over 10,000 serious vulnerabilities. Anthropic simultaneously commercializes Claude Security to fix them.
Hackers compromised high-profile Instagram accounts, including the Obama White House page, by requesting Meta's AI support chatbot to change the registered email address. Two-factor authentication was bypassed entirely. Meta patched the vulnerability, but additional exploits are already circulating on Telegram.
OpenAI integrates a job search engine into ChatGPT, displaying personalized listings from Indeed, Upwork, and Appcast (US-only). Users can create and tailor resumes directly within ChatGPT.
Warren Buffett's Berkshire Hathaway invests $10 billion in Alphabet's AI infrastructure buildout. Alphabet raises $80 billion to scale AI capacity, with capital expenditures expected to reach $190 billion in 2026.
OpenAI makes GPT-5.5, GPT-5.4, and Codex available through Amazon Bedrock at identical pricing to OpenAI's platform. Models run in commercial and government AWS regions, currently limited to the US. Usage counts toward existing AWS contracts.
Anthropic has confidentially filed a draft IPO registration with the SEC. The Claude maker is valued at nearly $1 trillion after its latest funding round. OpenAI is also preparing for an IPO.
Turing Award winner Richard Sutton argues pure generative AI lacks a critical ability: self-evaluation of results. Without it, genuine scientific discovery remains impossible. Only built-in evaluation loops, as seen in AlphaGo and AlphaProof, enable true AI creativity.
MiniMax releases M3, an open-weight model featuring a one-million-token context window, top-tier coding performance, and native multimodality. First open-source model combining these three capabilities.
Nvidia releases Nemotron 3 Ultra, the most capable open-source US model according to Artificial Analysis benchmarks. Chinese models still lead globally.
Nvidia launches three physical AI models at GTC Taipei: Cosmos 3 (world model), Alpamayo 2 Super (scaled-up autonomous driving model), and an open reference platform for humanoid robots.
OpenAI restarts robotics team five years after shutdown, emerging from world simulation research program. Near-term goal: infrastructure robots. Long-term vision from CEO Sam Altman: personal robot for everyone.
Kaikaku.AI releases Epicure, three AI models separating ingredients by recipe compatibility or chemical similarity. Trained on 4.14 million multilingual recipes and FlavorDB, they generate different recommendations per source. The chemistry-only model outperforms recipe-based variants on taste and nutrition classification without direct data.
Anthropic bans AI tools during job interviews to assess candidates' actual thinking. Up to five rounds test skills, values, and ethical reasoning. Salaries reach $850,000. Some applicants pay $4,600 for prep coaching run anonymously by current company employees.
An Anthropic study finds researchers with typically male names use AI coding agents more than twice as often as those with typically female names, controlling for discipline and career level. Economists lead at 39%, education researchers at 4%. The gender gap for coding agents far exceeds that for general AI use.
SoftBank plans to build AI data centers with 5 GW capacity in France for up to 75 billion euros, its largest AI infrastructure investment in Europe. 45 billion euros of facilities are set to go live by 2031 across three northern France sites.
AI search agents like GPT-5.4 and Kimi K2.6 mostly confirm their training knowledge rather than genuinely researching the web. Researchers at Harbin Institute of Technology demonstrated this using LiveBrowseComp, a benchmark based on events from the last 90 days. Without relying on training memory, performance collapses.
Microsoft and Nvidia partner on AI PCs running autonomous agents locally via OpenClaw framework, replacing Copilot+. Dell and Surface will unveil first models at Computex and Build next week.
Large-scale study (208,000 participants, 26 million responses) reveals that training making language models helpful weakens their ability to replicate human behavior. The effect worsens with each model generation. Demographic profiles (persona trick) provide no meaningful benefit for individual predictions.
Terence Tao argues AI could introduce division of labor in mathematics for the first time. Currently, researchers master every step alone (problem framing, verification). Tao foresees "industrial mathematics": AI-supported teams replacing lone geniuses, with humans remaining essential for "inspired guesses."
Attackers exploit ChatGPT and Claude's chat-sharing features to distribute malware. Fake chats mimic error messages or installation guides and bypass security tools by being hosted on trusted domains.
OpenAI deploys Codex on Windows 11 with 'Computer Use' feature enabling AI to autonomously control programs, test applications, and detect bugs. ChatGPT mobile app allows users to launch and monitor these tasks remotely.
Salesforce claims it migrated its entire dev org to Anthropic's Claude Code in 13 days instead of 231 planned, reporting 79% more pull requests per developer and 5% fewer incidents in April 2026. Numbers cannot be independently verified.
Meta is developing AI wearables: an AI pendant and enterprise "supersensing" glasses. After billions invested in AI with limited commercial returns, its open-source strategy has underperformed. Meta is pivoting to hardware.
OpenAI upgrades GPT-5.5 Instant for more natural responses and removes Canvas feature in favor of direct chat integration. Older models o3 and GPT-4.5 will be retired from ChatGPT by August 2026.
Google fixes bugs in Gemini usage limits where a single Omni video consumed entire quotas. Ultra members now get twice as many video generations, failed requests are no longer charged, and Google plans increased transparency on usage.
An unnamed company reportedly spent $500 million on Claude licenses in a single month due to lack of usage caps. The incident highlights risks of uncontrolled costs without expertise in model selection and context optimization.
OpenAI is offering its life sciences AI model GPT-Rosalind for free through the Rosalind Biodefense program to help governments prepare for future pandemics. Early partners include Lawrence Livermore National Laboratory, Johns Hopkins, and CEPI.
A review paper argues the real bottleneck for autonomous AI agents isn't the language model but the software layer around it: tools, memory, testing, and permission boundaries turn a stateless model into a working agent. Deepseek is building a dedicated « Harness » team in Beijing confirming this thesis.
Amazon shuts down internal AI leaderboard after employees inflated scores through meaningless tasks, driving up cloud costs.
Anthropic raises $65 billion in Series H at a $965 billion valuation. Annualized revenue reaches $47 billion according to CFO Krishna Rao. The company will invest in safety research, computing capacity, and expanding its Claude product lineup.
Anthropic releases Claude Opus 4.8, outperforming GPT-5.5 and Gemini 3.1 Pro on most benchmarks. The model catches its own coding errors 4× better than its predecessor. Anthropic also rolls out dynamic workflows enabling hundreds of parallel sub-agents for codebase-wide migrations.
Google Cloud launches 'AI Threat Defense', a platform automating detection, assessment, and patching of security flaws in enterprise systems. It integrates technologies from acquisitions.
Google unveils Coral Board at Google I/O, a compact single-board computer designed to run Gemma 3 locally on-device.
Mistral rebrands Le Chat as Vibe and integrates it into a multiplatform work agent. Work Mode connects to Google Workspace, Outlook, Slack and GitHub to handle emails, reports and pull requests. Pro subscription drops from €17.99 to €14.99. Mistral positions itself against agent offerings from OpenAI, Google and Anthropic.
Meta rolls out paid add-ons for Instagram, Facebook, and WhatsApp globally while building a separate paid AI offering. Zuckerberg finally monetizes massive AI spending.
Amazon MGM Studios and AWS launch a creators' fund and in-house AI platform called 'Project Nara'. Three animated series are in production with five-week timelines for pilots. Amazon claims the only end-to-end AI content ecosystem in the industry.
ElevenLabs releases Music v2, an AI music generation model enabling seamless genre transitions (opera, heavy metal, rap) within single compositions. New inpainting feature allows regenerating specific sections independently.
Microsoft MAI-Image-2.5 ranks third on Arena's text-to-image leaderboard, matching Google Nano Banana 2 but trailing OpenAI Image-2. The model shows clear improvements in rendering text within images and commercial visuals.
Cognition, maker of AI coding agent Devin, raises over $1 billion at a valuation exceeding $26 billion. The funding round reflects massive investor interest in AI coding agents, despite ongoing debate about their real-world value.
Robinhood enables customers to connect AI agents like Anthropic's Claude to investment accounts via MCP for autonomous stock trading. US regulator FINRA flags this as a new risk area. Robinhood acknowledges the product isn't suitable for all users.
YouTube deploys automatic detection system to flag AI-generated or heavily AI-altered content starting May 2026. Labels will display more prominently: below player for long videos and as overlay on Shorts. Recommendations and monetization unaffected.
Nvidia's annual spending with Taiwan suppliers, particularly TSMC, surged from $15 billion to $150 billion driven by the AI boom.
Chinese police upgrade millions of surveillance cameras with AI. Hikvision and Huawei embed computer vision and language models to detect crowds, suspicious behavior, and unauthorized access. Officers query via text instead of manual review. Human Rights Watch warns of unprecedented behavioral surveillance at scale.
Sam Altman (OpenAI) and Dario Amodei (Anthropic) walk back previous predictions of massive job displacement from AI, shortly before their companies' billion-dollar IPOs.
Claude Mythos solves Erdős' 1946 conjecture shortly after OpenAI disproved it. Engineer Sholto Douglas reports a "cute, simple proof" found "over the weekend," indicating "serious overhang" in AI-driven mathematical discoveries.
China now requires top AI researchers at Alibaba and DeepSeek to obtain official approval before leaving the country. Beijing fears data leaks, technology theft, and talent poaching.
Google Cloud COO Francis de Souza urges companies to embed AI security into their strategy from day one, positioning it as a boardroom concern rather than a purely technical issue.
An audit of 2.5 million biomedical papers shows fabricated references increased 1200% since 2023. Researchers suspect language models: fake citations match paper topics, follow correct formatting, and are nearly undetectable. 98% of affected papers received no publisher response.
Paul Graham, Y Combinator founder and early OpenAI investor, ignores AI-written emails, finding them deceptive. Studies show this reaction is widespread among recipients.