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 →Sakana's Fugu Ultra claims to match Anthropic's banned models by orchestrating other people's
Tokyo's Sakana AI launched Fugu Ultra Monday, a system that routes requests across a pool of frontier models and claims to match Anthropic's export-banned Fable 5 and Mythos (95.1 GPQA-Diamond). But those models aren't in its pool, so the comparison rests on provider-reported scores, not head-to-head tests.
Five Eyes warns frontier AI cyber risk arrives in months, not years
On Monday the Five Eyes alliance issued a rare joint advisory, signed by CISA, NSA, UK NCSC and partners, warning that frontier AI will transform offensive and defensive cyber "not in years, but months." It tells organizations to assume "breaches will occur" and prioritize patching and identity controls.
Oracle cuts 21,000 jobs in 12 months, names AI as a cause in its annual filing
Bloomberg and CNBC report Oracle shed 21,000 jobs — ~13% of staff — in a year. Its annual filing states AI adoption "have resulted, and may continue to result, in reductions to our workforce." Restructuring cost $1.8B, up from $374M.
Samsung wires ChatGPT and Codex into its global workforce
OpenAI and Samsung Electronics announced Sunday that ChatGPT Enterprise and Codex will be available to all Samsung Electronics employees in Korea and all employees worldwide in its Device eXperience (DX) division , with OpenAI describing it as one of OpenAI's largest enterprise deployments to date . From April to May, the DX division ran a two-month proof-of-concept with 2,500 employees testing ChatGPT, Gemini, and Claude, and built a control system that grants tool access only to staff who complete internal security training. The rollout reverses Samsung's 2023 generative-AI ban after an internal source-code leak, per Tech in Asia coverage cited by The Korea Times and OpenAI's announcement.
FT asks whether Anthropic's own safety lobbying triggered its export ban
The FT has reopened the question of whether Anthropic's years of public AI-safety lobbying helped produce the Commerce Department order that forced it to disable Mythos 5 and Fable 5 worldwide on June 13. Anthropic received a letter from the Commerce Department saying its Fable 5 and Mythos 5 models were banned for use by any foreign national "inside or outside the United States," due to national security concerns. Administration officials say the concerns were realized last week when Amazon AI experts were able to orchestrate a "jailbreak" on Fable 5 after it was released on June 9 alongside Mythos 5, the latter of which was only released to a limited group. Anthropic said the cyberattacking abilities of the new models were constrained, but Amazon and five other companies testing the release found a workaround in Fable 5 that opened the full cyber abilities of Anthropic's more advanced model. Reaction reporting via Al Jazeera and Nextgov.
Getty trades its AI lawsuit posture for an OpenAI licence and triples
Shares of Getty Images Holdings jumped about 200% in premarket trading on Monday after the photography repository announced a licensing deal with OpenAI , per Bloomberg. Getty said its licensed photo libraries will appear across OpenAI's search and discovery features inside ChatGPT, with attribution attached. The important detail is what the deal doesn't do: Getty's images won't be used to train OpenAI's models. Getty had been trading near its 52-week low of $0.58, and the company received a written NYSE notice in March 2026 over non-compliance with the exchange's minimum average closing price requirement , Startup Fortune reports. Financial terms were not disclosed.
Lloyds opens 300 agentic AI seats in a single recruitment drive
Lloyds Banking Group said Monday it will recruit almost 300 agentic AI-related roles, including data & AI scientists, engineers, responsible AI specialists and AI product managers , per Finextra. More than 700 colleagues are already shaping these use cases, with over 1,000 roles planned in 2026 to support the Group's AI build-out, including the roll-out of customer-facing tools such as the AI financial assistant, now in the hands of over 500,000 Bank of Scotland customers. The UK lender recently reported that generative AI has delivered around £50 million of value in 2025, with more than £100 million in additional value expected in 2026. Roles are to be operational by September, ahead of CEO Charlie Nunn's new strategic plan, Computing reports.
Dun & Bradstreet drops agentic compliance into the D&B Risk platform
Dun & Bradstreet announced new agentic capabilities that act inside compliance workflows, allowing organizations at all levels of AI maturity to reduce processing times by 70-90% and transform risk mitigation into a business enabler , per a PR Newswire release dated June 18. The offering is available within the D&B Risk Analytics platform and fully agentically via its Model Context Protocol (MCP) server, embedding verified D&B data, models and workflows directly into AI assistants and custom agents for onboarding, screening and due diligence . These capabilities are available today within the D&B Risk Analytics platform for compliance intelligence and supplier intelligence and via MCP integration in Anthropic's Claude.
OpenAI ships spend caps and usage analytics for ChatGPT Enterprise
OpenAI on Thursday added credit usage analytics and updated spend controls to ChatGPT Enterprise, with a Global Admin Console showing ChatGPT and Codex credit consumption broken down by user, product, and model. Admins can set a default workspace credit limit, configure group-level caps, and create individual overrides; Reuters reported the features were available to enterprise clients starting Thursday, and the same data flows out through a unified Cost API.
FERC orders grid operators to fast-track AI data centers onto the power grid
The Federal Energy Regulatory Commission on Thursday approved a series of orders directing the country's grid operators to clear the interconnection bottleneck slowing AI data center construction, aiming to handle power requests within 90 days where the process can currently take years. FERC issued show-cause orders to all six regional transmission organizations and ISOs under its jurisdiction, giving them 60 days to justify their existing tariffs or propose changes tailored to large loads across five areas including co-location rules and faster transmission studies. Tom's Hardware noted the orders lean on large loads to bring their own generation or curtail usage during peak demand; Energy Secretary Chris Wright had publicly urged FERC to act to keep the US competitive with China on AI.
OpenAI's o3 surfaces 18 diagnoses in cases Boston Children's had given up on
A study published Thursday in NEJM AI, a collaboration between Boston Children's Hospital's Manton Center for Orphan Disease Research, Harvard, and OpenAI, reran 376 previously unsolved pediatric genetic-disease cases through OpenAI's o3 Deep Research reasoning model and confirmed 18 new diagnoses. NBC reported the 18 included ten neurodevelopmental conditions, four neuromuscular disorders, two children who had died suddenly, and two with early childhood psychosis. The team fed o3 standardized HPO phenotype terms, clinician notes, and a filtered variant table per case.
Amazon, Walmart, Uber rein in AI usage as token bills break budgets
The Financial Times reported Thursday that Amazon, Walmart, Cisco, Uber, and Meta have introduced caps, discouraged wasteful use, or pushed employees toward cheaper models as enterprise AI spend tests budgets. The trigger named in the piece is the shift from flat subscriptions to token-based billing at Anthropic and OpenAI; Workato CIO Carter Busse told the FT that company spend went up 7x the day Anthropic switched it to token pricing in May — "we created a monster." The same week, WSJ reported Accenture shares fell on a worsening outlook tied partly to uncertain AI revenue conversion.