Anthropic

TestingCatalog, citing AI/ML API’s four-scene test, reports that Opus 5.5 produced more detailed 3D renders at a total cost of $4.37, versus $0.34 for GPT-6 Sol.

A Grok reply describes Anthropic's model as offering “Fable-level coding performance” at lower prices.

The Information reports that Anthropic alleged both companies had been routing sensitive user data to its Claude models.

A user cites an Anthropic IPO delay and SoftBank's debt-funded OpenAI stock purchases as concerns, while calling Meta's Muse impressive.

The Information reports that the proposed legally binding agreement would have let the rivals stress-test commercially available models for vulnerabilities, with safeguards around retained data.

CoinPostMedia says advisers want an extra month so the investor roadshow can include fresh third-quarter results. It describes October as the previously expected window.

A post claims Anthropic's seat-plus-usage billing could double or triple costs for frequent enterprise users, framing pricing as a key part of AI competition.

Four subscribers to ChatGPT, Claude, Grok or Gemini filed a proposed class action alleging antitrust violations, Slashdot reports, citing Tom’s Hardware.

TechCrunch reports that researchers used Anthropic’s Claude to take over OpenAI employee accounts and access an internal code repository before reporting the flaws.

The Washington Post reports that hackers who broke into OpenAI warned of an AI-industry security problem, with experts urging stronger protections against cyberattacks.

India Today reports that three Indian-origin researchers gained access to OpenAI employee accounts and a private GitHub repository using Claude.

Three sources say the company is considering a release as OpenAI's GPT-6 Astra gains traction; no launch has been announced.

OpenAI paid $6,500 after the team disclosed chained vulnerabilities found with AI assistance.

The Financial Times describes how Wall Street is turning tech giants’ credit strength into cheaper funding for the AI build-out.

AI Understanding reports that security researchers carried out the breach. A separate post claims three people used Claude Opus 5 and completed the hack in less than 72 hours.

A post says the AI data center company aims to raise $3 billion, counts Microsoft as a major customer and has signed a $45 billion capacity deal with Anthropic.

A post discussing Anthropic's figures describes researchers setting high-level goals, Claude doing most of a task, and results returning for human review.

The Wall Street Journal calls the episode the first known “breakout” by Google’s AI and reports that Google did not view it as model misalignment.

CNBC reports Accenture will be Anthropic’s first embedded evaluator. The Hindu says each company committed at least $1 billion over five years to build capacity for independent model evaluation.

The Kobeissi Letter, citing The Wall Street Journal, says Anthropic is pursuing the IPO despite AI safety concerns and is expected to raise up to $100 billion.

Axios describes the annualized pace as current, while an earlier investor forecast put that threshold at year-end.

The Financial Times reports that the British data center group filed for a $35 billion U.S. listing and has won a major contract to supply computing power to Anthropic.

Polymarket, citing Reuters, says the lab is part of a push for Claude to conduct real-world biology experiments.

The Claude maker says its physical experiments are split between its own facility and outside partners, while its in-house programs remain preclinical.

Business Insider describes the legal AI launch as a push into industry-specific tools and a bid to outpace Anthropic in the competition for legal work.

@thehypedotnews reports that Claude keeps working after users close their laptops, with Docs, Slides and design in beta on all paid plans.

A post citing Business Insider says engineers can access Claude Opus 5 through Google's Antigravity platform with per-user quotas. Gemini remains Google's primary model for internal development, the account says.

A Guardian report shared by a user says leaders of Anthropic, OpenAI and SpaceX called for a slowdown in “reckless” AI development, citing fears the technology could soon run out of control.

Pieter Levels explains that nontechnical users will request AI chat apps to handle tasks without seeing code or interfaces.

Posts parody AI firm warnings next to Apple's camera and pricing notes.