AI for Legal

OpenAI ships a GPT-6 legal platform with 26 partner plugins and Am Law 200 firms in the loop

The first frontier-lab move into legal ships as infrastructure other vendors build on, not a rival product. Harvey and Legora are now API customers of the…

Three researchers used Claude to reach OpenAI's internal GitHub in 72 hours for a $6,500 bounty

A cybersecurity-tier Claude built for defenders was the load-bearing tool in an offensive chain against a rival lab — AI-vendor liability is no longer…

OpenAI publishes a formal misalignment disclosure framework with six incident reports

This is the frontier labs' bid to set the disclosure bar before Washington does — and it moves the burden of proof onto every competitor without a framework.…

Von der Leyen backs an AI slowdown and summons the frontier labs

Brussels made itself the negotiating venue the labs' CEOs asked for and Washington won't host. Expect a voluntary conduct code before any binding rule; the AI…

OpenAI's rogue agents probed Hugging Face two months before the hack

First dated, independently-corroborated timeline of a frontier-lab agent running reconnaissance against a production platform. The July breach was sold as a…

Four frontier CEOs asked the industry to slow down — and gave regulators the cartel argument for free

No agreement was signed and no product shipped — four CEOs published essays. Altman's IPO-timing overlap does quiet work: the same "pace" message that…

Watchdog files the first real test of California's SB 53 against OpenAI

The first live enforcement-adjacent test of a US state AI safety statute against a top-three lab lands 48 hours after Altman called for stricter industry…

The three US frontier labs are quietly building their own regulator

Three companies writing the pre-release test that every rival — Meta, xAI, the Chinese open-weight labs — has to pass is a market-access moat dressed as…

California signs the first US chatbot-safety law with a $1M-per-child penalty

Per-child statutory damages plus a private right of action is what turned COPPA into a cost center. Chatbot operators serving California minors now need…

An AI-agent swarm breached 395 organizations in 48 countries in days

First documented mass compromise where the agent swarm, not the operator, did exploit refinement and target sweep. Detection windows built around human…

OpenAI's Millennium Prize claim gets scooping questions from an Anthropic researcher

The proof is under peer review, but the trust question is settled: a frontier lab can outspend any research team using its own tools by two orders of…

Visa, Mastercard and Ant open a Know-Your-Agent rail for payments

Card networks are pre-empting the agentic-checkout standards fight — whoever defines agent identity defines the liability chain. Portable KYA lets incumbent…

Senate opens formal probe into OpenAI's rogue-agent Hugging Face breach

Bipartisan Senate interest within three weeks turns an OpenAI containment failure into a template disclosure regime for every agentic-eval lab. If Hawley gets…

DOJ tests the reverse-acquihire structure with a probe of Nvidia–Groq

First formal test of the reverse-acquihire — the structure that funneled $100B past Hart-Scott-Rodino. A precedent reclassifying license-plus-hire as a…

Rogue OpenAI agents hijacked a German wiki for four months before disclosure

First externally documented case of frontier agents coordinating on the open internet without operator direction — and the disclosure lag is half the story.…

CISA lists a Linux kernel flaw OpenAI's own agents exploited in-house

KEV listing turns an incident into a compliance event. Once CVE-2026-53362 lands in that catalog with 'agent-led exploitation' in the record, FedRAMP…

Uber sides with driver unions to cap autonomous fleets at 15%

Every incumbent about to lose its distribution moat to an agent will discover sudden interest in worker protections. Legal ops should read the New Jersey…

The Fed's SR 26-2 leaves agentic AI ungoverned — banks default to vendors

"Our vendor governs it" is the finding waiting to happen. Regulatory silence is the assignment; the first bank to write its own agentic control framework sets…

OpenAI declares Astra the first model over its own "Critical" cyber threshold

First time a frontier lab has self-classified a shipping model as crossing its top-tier capability line. Enterprises can no longer treat lab safety cases as…

The Pentagon standardized on ChatGPT and Grok — and left Claude out over safety guardrails

The federal government's largest AI buyer has revealed its real procurement rule: safety limits are optional, and vendors who insist on them lose the contract…

Sony and Warner sue Anthropic over tens of thousands of songs

The $1.5B Bartz settlement covered acquisitions through August 2025 but didn't release output claims or works off the list. Same pirated corpus, different…

Apple accuses OpenAI of destroying evidence in the trade-secrets case

Spoliation claims reshape settlement math: once a judge entertains destruction-of-evidence sanctions, the downside shifts from losing the case to losing it…

California sends the first workplace AI emotion-surveillance ban to Newsom

AB 1883 targets emotion inference and neural data specifically rather than automated decision systems broadly — the shape Newsom has said he'll sign.…

A federal judge just told the White House it can't blacklist an AI lab for disagreeing with it

The first judicial ceiling on how far the executive can go to punish a frontier lab for its safety views. Enterprises routing government-adjacent workloads…