News & analysis, rated.
Every story from the daily brief, each rated so you know what matters. Hand-picked by the desk, not an automated firehose.
Browse the news archive →IDBS and Alchemi wire agents into a 21 CFR Part 11 audit trail for biopharma filings
IDBS, a Danaher company, and applied-AI firm Alchemi announced a partnership connecting Alchemi's agents to data in the IDBS Polar platform to draft CMC technical reports, clinical study reports, and submission dossiers. Each draft routes to a scientist for sign-off, with every human and AI action captured in a 21 CFR Part 11 audit trail. Alchemi says customer deployments have produced these documents up to 70% faster, with co-founder Anuj Chadha claiming agents draft submission-ready filings "in minutes".
Google's AMIE matches GPs on multi-visit disease management in a Nature study
Google Research published a peer-reviewed study in Nature on Wednesday showing AMIE, its medical reasoning agent built on Gemini, held its own against primary care physicians across simulated multi-visit cases. In a randomized, blinded virtual Objective Structured Clinical Examination (OSCE) study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multi-visit case scenarios designed to reflect UK NICE Guidance and BMJ Best Practice guidelines. Per Google's research blog, AMIE matched clinicians in overall management reasoning and scored significantly higher in plan preciseness and guideline alignment . The system pairs a dialogue agent with a separate management-reasoning agent that retrieves drug formularies and guidelines in long context.
An OpenAI–Molecule.one agent improved a real medicinal-chemistry reaction end-to-end
OpenAI and Polish biotech Molecule.one announced that a near-autonomous agent built on GPT-5.4 picked a research target, designed experiments, and ran them in Molecule.one's robotic lab, ultimately proposing an unexpected improvement to a workhorse oxidation reaction used in drug discovery. Paired with Molecule.one's Maria AI and specialized lab, the model proposed an unexpected way to improve a widely used reaction in drug discovery. The full process took about 2.5 months, plus another half month for human chemists to write up the results. The full account is on OpenAI's research index with a preprint attached.
JPMorgan pulls Claude from its Hong Kong approved-tools list
JPMorgan Chase removed Anthropic's Claude models from the internal list of approved LLMs available to its Hong Kong employees, Reuters reported Thursday, citing the FT. The wording of Anthropic's usage terms in its licensing agreement with JPMorgan prompted the bank to remove Claude models from an internal drop-down list of approved large language models available to employees in the Asian financial hub . The move follows a similar decision by Goldman Sachs, which in April removed Claude from a list of approved tools available to its Hong Kong-based bankers. The block lands days after the US Commerce Department ordered Anthropic to suspend exports of its Mythos and Fable models to all foreign nationals.
Anthropic lobbies G7 leaders for a US-led coalition while Trump negotiates its export ban
At Wednesday's G7 working lunch in Évian-les-Bains, Anthropic CEO Dario Amodei and Google DeepMind's Demis Hassabis pitched heads of state on a US-led AI coalition, per CNBC. Amodei said in his address to the group that areas of international cooperation should include structured access to frontier models, and trade of chips and critical components that excludes China, one person said. Amodei also said that countries should cooperate to address the risk of AI in cyber, bioterrorism and areas of intelligence, the person added. President Trump told reporters at the same summit that Anthropic negotiations are "going fine," days after the Commerce Department's export order took effect and Anthropic took Fable 5 and Mythos 5 offline globally.
Enterprise AI's price war is no longer hypothetical
The Wall Street Journal, via CNBC and Yahoo Finance, reported OpenAI is weighing significant token-price cuts in anticipation of similar moves from Anthropic, both companies running into hard enterprise budget ceilings. OpenAI is weighing major cuts to what it charges for tokens, the unit AI companies use to bill for their products, according to the Wall Street Journal. The move is a direct response to Anthropic, which OpenAI expects to lower its own rates. CEO Sam Altman has already acknowledged the problem publicly, saying AI costs have become "a huge issue" for corporate clients. Independent benchmark data from Artificial Analysis via CNBC put the gap in numbers: For each lab's most capable model: Anthropic's Claude came in at $4,811. OpenAI's ChatGPT: $3,357. DeepSeek: $1,071. Kimi: $948. Zhipu's GLM: $544. Claude is nearly nine times more expensive than the cheapest Chinese alternative for the same workload.
Stanford's AI Economic Indicators: early-career jobs track AI exposure, no economy-wide takeoff yet
Stanford's Digital Economy Lab launched the AI Economic Indicators last week, with its first research note updated this week. The note finds that for early-career workers aged 22 to 25, the two most exposed groups of occupations have seen noticeable declines since the introduction of ChatGPT, while the other three occupation groups have grown — and that the pattern becomes less stark, and ultimately disappears, for older workers. The lab's Takeoff Tracker monitors macroeconomic indicators associated with AI-driven economic change, and its current 12 indicators show no decisive evidence of takeoff at present. TIME's coverage of the underlying data put the relative decline for the most-exposed early-career workers at about 16%.
Z.ai's open-weights GLM-5.2 matches GPT-5.5 on long-horizon coding at one-sixth the cost
Z.ai (formerly Zhipu AI) released the full weights of GLM-5.2 on Hugging Face under an MIT license this week. On SWE-bench Pro, GLM-5.2 scored 62.1, beating GPT-5.5's 58.6 and its own predecessor GLM-5.1's 58.4. VentureBeat reports the model runs at roughly one-sixth the per-token cost of GPT-5.5. Independent confirmation arrived from Arena.ai: Code Arena, an Elo-style leaderboard built on blind pairwise human votes, placed GLM-5.2 at second overall with a score of 1,595 — first among models currently available, since Claude Fable 5 was removed following the US Commerce Department's June 12 export directive.
Nvidia's Jensen Huang says society needs "new social norms" for the AI age
In an AP interview Tuesday in Sherman, Texas, Nvidia CEO Jensen Huang argued AI adoption requires social adaptation on the scale of cars reshaping cities. "We need to create new social norms," Huang said, urging "everybody use AI" — comparing the shift to how the arrival of cars reshaped streets, sidewalks, and crosswalks. He also told the AP he supports government regulation and safety standards for AI and said national security should be a priority. Nvidia's market capitalization sits at roughly $5 trillion; OpenAI and Anthropic are potentially set to clear $1 trillion once publicly traded.
Agentic coworkers go GA: Microsoft Copilot Cowork and Databricks Genie One launch the same week
Microsoft moved Copilot Cowork to general availability on June 16. Billing for Copilot Cowork begins with general availability; tenants with at least one user in the Frontier program between March 30 and June 16 receive a grace period and will not be billed for Cowork usage until July 1, 2026. Nine partner plugins are live at launch — Enosix, Harvey, LSEG, Miro, Monday.com, Moody's, Morningstar, S&P Global Energy, and TeamsMaestro — with eight more planned from Adobe, Atlassian, Box, Canva, CB Insights, Databricks, MoneyForward, and Templafy. At GA, the service runs on Anthropic models including Opus 4.8 and Sonnet 4.6. The same day at Data + AI Summit, Databricks launched Genie One, an agentic coworker that does things on behalf of workers and reasons over both structured and unstructured data, including corporate information that lives outside the Databricks platform.
OpenAI Q1: revenue tripled to $5.7B, $3.7B burn, $73B on hand — IPO pressure eases
The Information reported that OpenAI burned through $3.7 billion in the first quarter, more than half its $5.7 billion in revenue, with both cash burn and revenue tripling from the same period the previous year. The company finished the quarter with more than $73 billion in cash and marketable securities, up from around $40 billion in December — most of that jump from fundraising rather than the underlying business — and has told investors it doesn't expect to turn profitable until the end of the decade. Reuters noted that if cash burn continues at the Q1 level, OpenAI will not need to raise additional funds in the near term, which could reduce pressure to pursue a public offering immediately.