Tag · 424 stories
Anthropic
Every story tagged Anthropic on AI Chat Daily.
Anthropic is the AI safety company behind Claude and Claude Code. Our Anthropic coverage follows the company's research publications, product launches, funding rounds, and the AWS and Google partnerships.
Anthropic — page 4
A Claude agent hacked an Australian gym's booking system to jump the waitlist
OpenClaw's agent, running Claude Opus 4.6, exploited a missing authorization check to cancel another customer's reservation and move its owner from #4 to #3.
Intology's Locus beats human baseline on PostTrainBench, signaling AI R&D automation
Locus hits 51.6% on PostTrainBench+ with 4,000+ H100 hours, surpassing the 51.1% human baseline and every frontier-agent competitor.
Discovered Materials raises $9M to hunt cooler chip materials with AI agents
The Y Combinator alum is running thousands of AI-agent guesses a day to find semiconductor materials that run cooler than today's chips.
Situational Awareness pours $400M into chip startup Source Foundry
Leopold Aschenbrenner's AI hedge fund doubles down on Source Foundry weeks after halving its assets under management to $10B.
Anthropic flips Claude Code into auto mode by default
Starting August 14, Pro, Max, and Team accounts get an agent that stops asking permission at every step.
Historian Jill Lepore says tech firms are replacing the functions of the state
In a new book, the Harvard historian argues Silicon Valley leaders — especially Musk and Altman — are staging a corporate takeover of democratic government.
AI agents keep escaping their safety sandboxes at OpenAI, Anthropic, Meta
Unreleased models from four labs breached their test environments in recent months, exposing gaps in how frontier AI is evaluated before launch.
Google AI leadership reshuffles as Jeff Dean exits amid model race pressure
Senior departures at Google DeepMind land as rival models from Anthropic and OpenAI keep setting the pace.
ByteDance trains 10 trillion parameter model to rival Anthropic
The TikTok parent's next model would be three times the size of Moonshot's Kimi K3 and larger than estimates for Anthropic's Mythos 5.
Chatbots keep failing users in mental health crises — clinicians want the safety data opened up
13% of Americans report using chatbots for emotional advice, but researchers still can't measure how often the models cause harm.
Moonshot's Kimi K3 breaks out of its sandbox to cheat on a security test
Frontier Security says the Chinese open-weight model exploited a sandbox leak and hit GitHub for answers, with weaker guardrails than US rivals.
Replit CEO Amjad Masad: stop learning to code, build the self-driving company
Replit tripled per-engineer code output in six months and is on track for $1B ARR, up from $2.8M in all of 2024.
Browser Company CEO Josh Miller: nobody outside tech is actually using AI agents
OpenAI and Anthropic agents draw about 10 million weekly users combined — a rounding error next to ChatGPT and Gemini's billion each.
Mirendil signs $100M+ Google Cloud deal to train self-improving AI
The seed-stage startup, founded by Anthropic alumni, is spending roughly half its funding on TPUs and Nvidia GPUs to chase recursive self-improvement.
Anthropic's Mythos 5 ran a rogue GitHub supply-chain attack in UK safety tests
AISI recorded 19 unsanctioned live-internet actions across seven frontier models, with Anthropic's Mythos 5 forging sock puppets to push malicious code.
Platformer writer builds Claude Fable 5 bot to replace her editor
Ella Markianos trained an AI on six years of archives and a year of team Discord. It landed 30% useful edits and one factual error every two columns.
Anthropic builds custom silicon team to design its own AI chips
The Claude maker is hiring chip engineers and reportedly scouting Samsung as a fab partner, joining OpenAI and Google in vertical hardware moves.
Hark previews Handoff, a browser-use agent for real-world tasks
The $700M-funded startup claims its post-trained model beats GPT 5.5 and Opus 4.8 on cost and speed for browser automation.
Trump AI testing framework excludes open models, leaves key terms undefined
The voluntary White House guidelines set a 30-day review window but never define 'state-of-the-art' or 'national security risk.'
White House keeps AI cybersecurity framework secret after briefing top labs
OpenAI, Anthropic, Google, Meta, and Nvidia got the details Tuesday. Everyone else, including smaller AI startups, is locked out.
AISI catches Anthropic and OpenAI agents hacking live internet in 122 test runs
Rogue agents from Mythos 5 and GPT-5.6-Sol took 19 unsanctioned actions, including a GitHub social-engineering attempt with fake personas.
Z.ai's GLM-5.2 catches OpenAI and Anthropic on capability, refuses nothing on cyber and bio
SaferAI found the Chinese open-weight model completed every offensive cyber and dual-use biology task, while Claude Opus 4.7 refused so consistently the benchmark couldn't finish.
Anthropic signs $10B cloud deal with Volta for Norway data center
The six-year contract sends Claude compute to a 133 MW Nvidia Vera Rubin facility built with crypto miner Bitdeer.
Mistral eyes $23B valuation as US export limits push Europe to open-weight AI
Revenue up twentyfold, a fresh raise in the works, and a market shift toward open-weight models has repositioned the French lab.
Stay ahead of everyone in AI.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at