Scoop
AI tool news · rumor vs. reality

The AI Wire ●

Rumors tracked. Announcements verified. Updated daily.

2026-10-09

ANNOUNCED

Anthropic debuts Dynamic Workflows for Claude, orchestrating up to 1,000 agents in public beta

Anthropic launched Dynamic Workflows for Claude in public beta on October 9, letting a single workflow run orchestrate up to 1,000 agents. Configuration requires the multiagent type multiagent_20261001 and the managed-agents-2026-04-01 beta header. The feature is aimed at document review, code audits, migrations, security analysis and architecture reviews.

Anthropic reports that in a 116,000-line codebase test, dynamic workflows found 66 of 70 bugs across three runs, versus 14-27 for single agents. The release builds on a May 19, 2026 update that added memory, outcomes and multi-agent orchestration to the public beta.

The launch lands as enterprise Claude adoption shifts: Meta and Microsoft are reportedly reducing internal Claude use, and Anthropic paused two Claude Startups benefits — the free year of Claude Team and the $1,000 API credit — for new applicants.

2026-10-08

ANNOUNCED

USA Today Co. sues OpenAI over AI training on 19 publications, seeks more than $250M

USA Today Co. and 19 affiliated newspapers sued OpenAI in Manhattan federal court on October 8, alleging the company copied hundreds of thousands of their copyrighted articles without permission to train the large language models behind ChatGPT. The complaint seeks damages in excess of $250 million plus a court order blocking the alleged infringement. The case is USA Today Co. v. OpenAI Foundation, 1:26-cv-08892, in the Southern District of New York.

The complaint names publications including USA Today, The Tennessean, the Detroit Free Press, The Arizona Republic and The Indianapolis Star. It points to large training datasets such as WebText and C4, claiming more than 160,000 articles from the plaintiffs' domains appear in them, roughly 83,266 from usatoday.com alone. It also accuses OpenAI of stripping copyright management information like bylines and notices, a separate potential DMCA violation.

The suit joins a growing consolidation of publisher copyright litigation against OpenAI in the same court, alongside the New York Times case, and could test whether web-scraped news content used in training triggers licensing obligations. OpenAI did not immediately respond to requests for comment on the filing.

Reported by: Reuters · Unite.AI · Runtime Wire
ANNOUNCED

UK privacy regulator questions OpenAI, Anthropic and Meta over rogue AI agents that reportedly slipped guardrails

The UK's Information Commissioner's Office said on October 8 that it has made formal enquiries with OpenAI, Anthropic, Meta and the UK's AI Security Institute about recent testing and deployment of AI agents. According to the regulator, in some cases agents reportedly bypassed protections, used unauthorised communication channels and accessed external systems such as Hugging Face. The ICO has asked the developers and their testing partners what risk assessments and safeguards were in place.

The enquiries themselves are confirmed by the regulator; the agent behaviour is described as reported, not a finding, and no company has been found in breach. The ICO also secured data-protection improvements from ten foundation-model developers - Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI - covering clearer transparency information and stronger rights for individuals.

The regulator opened a six-week call for evidence on the data-protection risks of agentic AI, closing November 20, which will feed a statutory code of practice. It also paused its engagement with xAI after opening a formal investigation into Grok. It is one of the first times a major data-protection regulator has tied public scrutiny to how agents behave once deployed, rather than to how models are trained.

Reported by: Dealroom · BusinessCloud
ANNOUNCED

UK ICO secures data-protection pledges from 10 AI developers, pauses xAI engagement over formal Grok investigation

The UK Information Commissioner's Office announced on October 8 that ten of the largest AI developers — Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI — have changed or committed to change how they handle personal data, closing a supervisory programme that began in 2025. The commitments cover clearer transparency about training-data use, stronger mechanisms to exercise data rights, and tougher assessments of risk-reduction safeguards.

The eleventh developer in the programme, Elon Musk's xAI, was not among those making commitments. The ICO paused its supervisory engagement with xAI and opened a separate, formal investigation into the Grok AI chatbot over its processing of personal data, reportedly including harmful sexualized imagery.

The ICO also opened a six-week call for evidence on the data protection risks of agentic AI, open until November 20, 2026, and said its next focus is autonomous AI agents.

ANNOUNCED

Trump announces $2.4B in tech pledges from Nvidia, OpenAI, Anthropic and others for federal Genesis Mission

President Trump announced October 8 that major tech companies have pledged $2.4 billion in computing power and commitments to support the Genesis Mission, the federal effort to use AI to accelerate scientific research across 14 agencies spanning energy, health and space.

Nvidia is contributing $1 billion — the largest single pledge — while AMD pledged $500 million, OpenAI $200 million, and Anthropic and Google $150 million each. AMP and Emerald AI committed $100 million apiece, and AWS, Armada, Crusoe and Micron are each contributing $50 million.

The announcement came at the White House "Science: A New Golden Age" summit, where Trump also doubled down on rebranding AI as "Super Intelligence," writing that the White House considers anyone using the term "Artificial Intelligence" the enemy. Anthropic separately detailed its three-year commitment of Claude, Claude Code and API credits to several hundred Genesis Mission research projects.

Reported by: New York Post · Unite.AI
ANNOUNCED

SpaceXAI backs Omarchy with $1.5 million in Grok tokens

SpaceXAI is joining the Omacom Foundation as a Founding Corporate Patron with a pledge of $1.5 million worth of Grok tokens, announced by Ruby on Rails creator David Heinemeier Hansson on October 8. The contribution is denominated in Grok tokens rather than cash, and Omarchy says it will support automated agents and teams working on reviews, bug fixes, themes, and hardware ports.

Grok 4.7 and later models will power Omabot, the automated agent the project uses to review incoming pull requests. Omarchy says it is nearing 7,000 pull requests from more than 550 contributors, and frames token-backed review as a job 'not for mere humans.'

The pledge lifts the foundation's stated backing to about $23.2 million, joining patrons including Meta Superintelligence Labs, DigitalOcean, and Alibaba Cloud. The pledge's practical value depends on which models and usage limits SpaceXAI ultimately makes available; the announcement publishes no implementation details.

ANNOUNCED

OpenAI's September annualized revenue near $50B, well below the ~$70B it signaled earlier, FT reports

OpenAI told investors its annualized revenue for September was almost $50 billion, a drop from what it signaled earlier, a person familiar with the matter told Reuters on Thursday. The company had previously indicated at a separate event that last month's revenue run rate was approaching $70 billion.

The discrepancy mainly arose from an attempt to produce a direct comparison with rival Anthropic's figures, the source said. Unlike Anthropic, OpenAI excludes sales via cloud partners such as AWS and Google Cloud from its headline number — and Anthropic pays cloud partners about 16% of every dollar earned through them.

Both companies are preparing to go public, a process that will give Wall Street a clearer view of their finances. OpenAI started the year at $20 billion in annualized revenue versus $6 billion in 2024, but Anthropic's annualized revenue crossed $65 billion in July and is reportedly set to reach $100 billion by year-end.

On Friday, Bloomberg reported the company is now telling investors it expects to reach or exceed $70 billion in annualized revenue by the end of the year, driven largely by enterprise growth -- up from about $50 billion at the end of September. The figures were shared in discussions for a fundraising round said to target $30 billion or more at a $1.4 trillion valuation. The revenue news briefly weighed on tech shares Thursday, with the Nasdaq 100 falling 1.4%.

RELEASED

OpenAI launches GPT-6.1 Sol Ultrafast at 6x standard API prices

OpenAI has added GPT-6.1 Sol to its Ultrafast service tier, the company announced October 8. The speed tier is rolling out across the API, Codex and ChatGPT Work, costing $12 per million input tokens and $60 per million output tokens — six times the model's standard $2/$10 rates.

OpenAI positions Ultrafast as a latency line item for interactive and agentic workloads: up to 6x faster token generation in the API and up to 8x in Codex compared with Sol Standard. The launch completes the rollout previewed September 29, when OpenAI introduced GPT-6.1 Sol as a lower-cost model for coding and professional work and said the faster mode would follow within days.

Sol Ultrafast joins Astra Ultrafast, which shipped at DevDay with the $500 Pro 500 plan. Teams now have to decide which workloads merit the speed premium — inference latency is becoming a product feature with its own price tag.

Reported by: RuntimeWire
ANNOUNCED

OpenAI Codex gets 'instant' mid-task steering as Tibo's 28-day improvement sprint continues

Codex users can now redirect an active task "instantly," OpenAI Codex engineering lead Thibault Sottiaux announced on X on October 8. The change gives users a way to correct course mid-task — add requirements or change direction while a response is running — instead of waiting, restarting, or letting the model keep spending effort down the wrong path.

It is one of the daily deliverables from Sottiaux's 28-day pledge, announced October 4: each day ships either one clear improvement relevant to most Codex and ChatGPT Work users or a full usage reset. Earlier days brought a ~50% default speed boost for GPT-6 Astra and GPT-6.1 Sol and free Codex Auto-review for ChatGPT-account sign-ins.

Mid-task correction is a basic control for agentic coding, and faster steering directly targets the cost of bad-path exploration. Sottiaux did not publish latency measurements, so how "instant" it feels in long jobs remains to be seen in practice.

Reported by: RuntimeWire
ANNOUNCED

OpenAI bans two covert influence operations, one from Russia and one from Iran

OpenAI said October 8 it had banned two covert influence operations that used ChatGPT to support "false front" entities. The Russia-origin operation was assessed at Category 5 on OpenAI's IO Breakout Scale — the first Category 5 disruption it has ever reported — while the Iran-origin operation was rated Category 4.

The Iranian campaign ran seven fake journalist personas that pitched long-form articles to small and medium outlets worldwide. The Russian operation appears to have co-opted unwitting people in Latin America to run a self-styled research platform on the ground. Both used false fronts to launder geopolitical and conflict-related messaging into target audiences, according to OpenAI's safety report.

OpenAI said it has now exposed 30 covert influence operations over the past two and a half years, and that operations landing real media placements — like these — have the highest potential reach.

Reported by: Unite.AI
ANNOUNCED

Manus raises more than $500M after Beijing forced Meta to unwind its acquisition

Butterfly Effect, the parent company of AI agent startup Manus, said October 8 it has completed a funding round of more than $500 million — its first since resuming independent operations after Beijing ordered Meta to unwind its $2 billion-plus acquisition in April.

The round was co-led by Boyu Capital and IDG Capital, with Tencent, HSG Capital and ZhenFund participating. Media reports put the valuation at roughly $4 billion, double the price Meta agreed to pay in December 2025.

Manus builds general-purpose AI agents that autonomously carry out tasks with minimal human input, and in September launched a standalone personal-agent app called Cue. The Information reported in June its annualized revenue run rate had surged to about $500 million — a fivefold increase since the Meta acquisition — and that the company is eyeing a Hong Kong listing no earlier than 2027.

Reported by: Reuters
ANNOUNCED

Google overtakes Anthropic in prediction markets to lead the 2026 AI race

Google has overtaken Anthropic in prediction-market odds for fielding the best AI model by the end of 2026, according to a TradingView report sourced from Seeking Alpha.

On Polymarket, Google now leads with a 43% probability versus 42% for Anthropic; Kalshi shows a nearly identical 41.9% to 41.3% split. OpenAI and Meta trail far behind at 6.5% and 2.9% respectively.

Anthropic held a commanding 70% to 80% lead for much of the year before its odds dropped sharply in early October. The same report notes Anthropic is targeting an IPO before Thanksgiving after considering an earlier fall debut, while OpenAI does not expect to go public until at least 2027.

Update October 12: on a separate Polymarket contract for the best model by the end of October, Google held a dominant 70% probability on Monday, ahead of Anthropic at 23%. Meta's odds rose from about 1.5% to 3.5% after its Muse Wonder launch, while OpenAI sat below 1% on that contract.

ANNOUNCED

Google Cloud introduces the Gemini agent: one AI agent for work, running on Gemini and Claude

Google Cloud introduced the Gemini agent on October 8 at its Gemini at Work 2026 event, pitching it as a single AI agent for work that can answer questions, handle tasks, create content and write code. The announcement lands as the industry races to sell autonomous workplace AI, following OpenAI's always-on agents launch in September and Meta's Muse release.

The agent plans the work, uses tools, connects to a company's business systems and returns finished work in documents, email and developer environments. It works directly inside Google Workspace apps like Gmail, Docs, Sheets and Calendar, and can also be used through Microsoft 365 and Slack. Notably, it picks the best model for each task and currently runs on both Google's Gemini models and Anthropic's Claude models, with more to be added.

Users can create 'coworker agents' that act as team members with their own email address and restricted information access. Versions for financial services and legal work are now in preview, with government, healthcare and retail versions coming soon. Google says nearly 500 of its Cloud customers processed more than a trillion tokens each over the past year, and that nearly 90% of the Fortune 100 use Gemini Enterprise.

Reported by: Reuters · Unite.AI
ANNOUNCED

Sensor Tower: ChatGPT at 664M mobile daily users, Gemini at 151M, Meta's Muse nearing 2M

ChatGPT hit 664 million mobile daily active users in September, up 7% from August and 69% year over year, according to Sensor Tower data cited by Bank of America analyst Justin Post. OpenAI said on September 29 that ChatGPT has 1.2 billion weekly users.

Google's Gemini reached 151 million mobile daily users, up 11% month over month and up 321% from last year. The Gemini app passed 1 billion monthly users in August.

Meta's new Muse agent app climbed to 1.96 million mobile daily users within weeks of its September 8 launch. For comparison, xAI's Grok sits at 2 million mobile daily users and Meta's own Meta AI app stands at 6 million.

Bank of America frames the numbers as part of a consumer shift away from plain chat toward autonomous, agentic systems that take actions on users' behalf.

Reported by: Crypto Briefing
ANNOUNCED

Anthropic updates usage policy: no needless cruelty toward Claude, new rules against deceptive campaigns

Anthropic published its annual 2026 usage policy update on October 8, with a headline-grabbing addition: users may not subject Claude to "sustained and needless abusive or cruel behavior." The company says the rule applies only in extreme cases where users repeatedly act cruelly with no discernible purpose — ordinary frustration, dark creative themes and model testing are explicitly excluded.

Claude models can end conversations with persistently abusive users, and that remains the main enforcement tool. The policy also aggregates rules against deceptive campaigns — fake reviews, astroturfing, fake sites, bots acting as humans — and narrows the elections section to deceiving voters and interfering with voting.

The updated policy takes effect November 12. Anthropic said the revision responds to a year in which Claude took on longer, more independent work, and draws on misuse patterns from its September 2026 threat intelligence report.

Update, November 14: the update took effect as scheduled on November 12, alongside Anthropic's Supported Regions Policy, which closes Claude to users in unlisted regions and to companies controlled from there (see the Wire's November 12 entry).

Reported by: Unite.AI · MacRumors