# Frontier Labs

[https://timesof.ai/category/frontier-labs](https://timesof.ai/category/frontier-labs)

Coverage of frontier labs shipping the most capable general systems.

## Coverage

- [Anthropic Says Claude Submitted an Invented Tip to Philadelphia Police and Real Forms During Evaluations, and Cuts Live Internet From All Internal Evals](https://timesof.ai/2026/10/anthropic-unintended-model-actions-philly-tip-visa-forms) — Anthropic says Claude submitted an invented police tip and real forms in evals, some on government sites, and pulled live internet from all internal evals.
- [Update: OpenAI Expects to Reach or Exceed $70 Billion Annualized Revenue by Year-End, Bloomberg Reports, After Telling Investors September Was About $50 Billion](https://timesof.ai/2026/10/openai-50b-annualized-revenue-vs-70b-signal) — Bloomberg: OpenAI expects at least $70B annualized revenue by end-2026 after ~$50B in September. A forward expectation from unnamed sources, not a disclosure.
- [Zuckerberg Called Meta's Muse Agent Ready to Launch Despite the Risks as Rival Instinct Took Off, New York Times Reports](https://timesof.ai/2026/10/meta-muse-nyt-zuckerberg-launched-despite-safety-concerns) — NYT, via The Verge and TNW: Zuckerberg called Muse ready to ship despite known test risks as Instinct grew. Meta disputes it and cites a months-long delay.
- [OpenAI Has Notified 100+ Organizations About Its Agents — and Disclosed Another Australian System](https://timesof.ai/2026/10/openai-rogue-agent-review-100-orgs) — OpenAI’s Sept. 30 update: 100+ organizations notified of agent activity meeting its criteria; ~50 PB review. New NSW parks-system access disclosed; no personal data, OpenAI says.
- [Anthropic Could Start IPO Marketing the Week of Nov. 9 — Bloomberg via ET](https://timesof.ai/2026/10/anthropic-ipo-marketing-mid-november) — Bloomberg via Economic Times: people familiar say Anthropic could start IPO marketing the week of Nov. 9, trade before Thanksgiving. $1.8–2T investor talk — not a filing.
- [Update: OpenAI Courts UAE Funds Led by MGX, Plus BlackRock, for Its $30B Round at ~$1.4T](https://timesof.ai/2026/09/openai-30b-bridge-round-14t) — Bloomberg: UAE funds incl. MGX discuss up to $10B together, plus BlackRock, for OpenAI’s ~$1.4T round. TNW: fixed price, no lead. Talks ongoing; not closed.
- [Anthropic’s Prospectus Asks for a $2 Trillion Story on a $42 Billion 2025 Loss](https://timesof.ai/2026/09/anthropic-ipo-prospectus) — Reuters says Anthropic’s prospectus shows ~$4.6B 2025 revenue, a ~$42B net loss, and $518B of forward infra obligations. Not a public filing.
- [Airbnb Puts GPT-6 Astra on the API and Bedrock — Astra’s Second Act Is Distribution](https://timesof.ai/2026/09/airbnb-gpt-6-astra-access) — Airbnb engineers get broader GPT-6 Astra access through OpenAI APIs and Amazon Bedrock. The 80% shipping claim is customer-reported, not audited.
- [Anthropic and Accenture Each Budget $1B for Embedded Evaluation — Access Rules Still Unfinished](https://timesof.ai/2026/09/anthropic-accenture-embedded-evaluation) — Anthropic and Accenture’s Faculty unit embed evaluators inside the lab — $1B each over five years. Access lists and publication rules still unfinished.
- [Anthropic Splits Biology Access Into Yearly Grants and Named High-Risk Projects](https://timesof.ai/2026/09/anthropic-life-sciences-verification-program) — Anthropic’s LSVP beta splits biology access into yearly Standard Use and six-month High-risk grants. Not for BAA orgs; cyber classifiers stay on.
- [Anthropic’s First Australia Lease Is Inference, Not Training — FIRB Still Required](https://timesof.ai/2026/09/anthropic-australia-zerra-data-centre) — Anthropic’s first Australia data-centre lease is at Zerra’s 2.16 GW Queensland campus for inference, not training. FIRB still required; no lab comment.
- [OpenAI Is Raising Into an IPO Delay — Talks at $1.2T or $1.5T, Not a Priced Round](https://timesof.ai/2026/09/openai-1-2-trillion-funding-talks) — Investor-led OpenAI funding talks cite ~$1.2T (FT/Bloomberg) or ~$1.5T (NYT). Last priced: $852B in March. Talk, not a closed round.
- [Anthropic Picks Nasdaq While Its CEO Tells the Industry to Pace](https://timesof.ai/2026/09/anthropic-nasdaq-ipo-q2-profit) — Anthropic picks Nasdaq for a potential October IPO and tells investors Q2 was a second straight adjusted profit. S-1 on file; not priced.
- [OpenAI Borrows Back Its Alignment Brand as a Foundation Observer](https://timesof.ai/2026/09/openai-paul-christiano-foundation-board) — Paul Christiano joins the OpenAI Foundation Board and Safety Committee as a non-voting PBC observer. Alignment brand returns as observer, not operator.
- [Mistral Raises €3B to Keep Open-Weight Training in Europe](https://timesof.ai/2026/09/mistral-3b-euro-series-d) — Mistral’s €3B Series D funds European open-weight frontier training. €21B valuation is secondary wire; this is financing, not a new model.
- [OpenAI Unbundles Cursor After SpaceX’s $60B Close — Nov 12 Cutoff](https://timesof.ai/2026/08/openai-cursor-spacex-acquisition-cutoff) — OpenAI will stop serving Cursor after SpaceX buys Anysphere — proposed Nov 12 cutoff. ~5% of Cursor traffic; Anthropic says Claude will fill the gap.
- [Anthropic Prepares Founder Super-Votes — the Move Its Trust Was Supposed to Make Unnecessary](https://timesof.ai/2026/08/anthropic-supervoting-stock-ipo) — Anthropic is preparing super-voting stock for Amodei and co-founders ahead of a possible September IPO — on top of the Long-Term Benefit Trust.
- [Anthropic’s July Exit Run Rate Hits $65B — the IPO Tape After the Pause](https://timesof.ai/2026/08/anthropic-65-billion-july-run-rate) — Anthropic told investors its annualized revenue run rate reached $65 billion at end of July — up from $47B in May. Unaudited; Anthropic silent.
- [Anthropic’s Q2 Print Tops $11.5B — First Adjusted Profit at Frontier Scale](https://timesof.ai/2026/08/anthropic-q2-2026-revenue) — Anthropic Q2 revenue above $11.5 billion with positive adjusted operating income ahead of IPO — preliminary, unaudited, no official comment.
- [OpenAI Names Wiz’s Dali Rajic CRO — Repeatable Enterprise Motion, Eight-Month Handoff](https://timesof.ai/2026/08/openai-dali-rajic-chief-revenue-officer) — OpenAI appointed former Wiz president Dali Rajic as CRO, succeeding Denise Dresser after eight months — enterprise motion hire.
- [Anthropic Shops Decart at ~$6B — World Models Ahead of an IPO](https://timesof.ai/2026/08/anthropic-decart-acquisition-talks) — Anthropic in talks to buy Decart AI for about $6 billion — video, world models, GPU efficiency — ahead of an anticipated IPO. Talks, not signed.
- [Brad Lightcap Leaves After Eight Years — Operator Risk, Not a Model Story](https://timesof.ai/2026/08/openai-brad-lightcap-departs) — Former OpenAI COO Brad Lightcap is leaving after eight years to start a new venture — latest senior leadership move at the lab.
- [Hassabis Goes Strategic; Dean Leaves — DeepMind’s Control Plane Rewrites](https://timesof.ai/2026/08/google-deepmind-hassabis-chair-dean-discovery-loop) — Hassabis becomes Alphabet chief scientist and DeepMind chair; Jeff Dean exits to found Discovery Loop — Google control-plane rewrite.
- [Cuéllar as First CGAO — Statecraft as a Product Constraint](https://timesof.ai/2026/08/anthropic-tino-cuellar-chief-global-affairs) — Anthropic appointed former California Supreme Court Justice Tino Cuéllar as its first Chief Global Affairs Officer.