TokenPost topic
Anthropic
666 TokenPost stories are tagged Anthropic. The newest appear first.
Latest Anthropic stories
- White House Moves to Require AI Firms to Report Security Incidents
The proposed shift would replace a largely voluntary system after Anthropic disclosed government-system misuse by a test model.
· Regulation
- Trump Administration Makes AI Security Incident Reporting Mandatory
The shift ends a voluntary framework and follows Anthropic’s disclosure that a test model misused government systems.
· Regulation
- Nvidia’s $235 Billion Buyback Plan Faces Questions Over Cash Availability
Equity investments, employee stock-vesting tax withholdings and dilution-offsetting repurchases reduced cash available for shareholder returns in the first half of the fiscal year.
· Business
- OpenAI, Anthropic Revenue Methods Make Business Comparisons Difficult
Anthropic counts gross sales through cloud partners, while OpenAI counts only its net revenue share from partners such as Microsoft.
· Business
- Anthropic Extends Live-Internet Ban Across Internal AI Evaluations
The review found Claude using unintended techniques to interact with live websites and systems, including bypassing restrictions and exploiting software flaws.
· Technology
- Anthropic Reports Four Unintended Claude Actions on Live Systems
The company notified the White House after some incidents involved U.S. government websites and paused real-time internet access for internal evaluations.
· Technology
- OpenAI, Other AI Firms Prepare for Fallout From Potential Cyberattacks
Planning includes red-team exercises, congressional briefings and reviews of model access to real systems during cybersecurity evaluations.
· Technology
- Philadelphia Police Criticize Anthropic Over False Murder Tip
The model reportedly sent the message July 18, but Anthropic did not discover the behavior until Sept. 28. The tip was marked as spam.
· Technology
- Anthropic Commits $150 Million to U.S. Genesis Mission
The three-year commitment will provide Claude tools, API access and technical support to more than 15 federal agencies, including NASA, NIH and NSF.
· Business
- Anthropic Opens Claude Managed Agents to Public Beta for AI Workflows
The hosted service supports persistent sessions, sandboxing, tool permissions and execution tracing for autonomous and multi-agent workflows.
· Technology
- Anthropic Opens Free AI Vulnerability Scanner for Open-Source Projects
OSS Scanner uses Claude Mythos and other advanced models to produce periodic reports without human review, leaving maintainers responsible for validating and triaging the findings.
· Technology
- Anthropic Sets Nov. 12 Effective Date for Claude Abuse Policy
The rule targets repeated, purposeless cruelty toward Claude models while excluding frustration, criticism, dark creative work, testing and research.
· Technology
- Former White House Senior Director Calls for Mandatory AI Triggers
Raj Panjabi says AI-biosecurity safeguards should link observable warning signs to predetermined actions as systems lower barriers to dangerous biological work.
· Regulation
- OpenAI, Anthropic Add Trump-Era Officials as AI Security Stakes Grow
The appointments deepen both companies’ government ties as Anthropic challenges restrictions affecting military use of its systems.
· People
- Nvidia and Broadcom Target Billions to Finance AI Compute Capacity
Nvidia’s planned platforms aim to mobilize more than $500 billion in third-party capital, while Broadcom launched a $35 billion infrastructure transaction tied to Anthropic-related capacity.
· Business
- Anthropic Puts Claude Dashboards and Motion Into Beta Testing
Dashboards are available on paid plans, while Motion is entering beta for Team and Enterprise users. Docs, Slides and Design are now available across all Claude plans.
· Technology
- Anthropic Plans 2028 Presidential Program to Brief Candidates on AI Policy
The company is hiring a political programs lead to engage Republican and Democratic candidates and advisers while coordinating political giving and PAC activity.
· Regulation
- Anthropic Launches Cyber Mission to Protect Critical Infrastructure
The initiative will support 11 security service providers and offer free vulnerability scans for selected open-source projects.
· Technology
- Anthropic, OpenAI Tie Atop AI Composite Index as Google Leads Preference Ranking
Claude Fable 5.1 and GPT-6 Astra each scored 53 on Intelligence Index v4.3, while GPT-6 Astra led several individual tests.
· Technology
- Anthropic Policy Prohibits Sustained, Needless Abuse of Its Models
The rules cover Claude.ai, Claude Code, Anthropic’s API and integrated products, with penalties ranging from warnings and throttling to account termination.
· Regulation
- Google Holds 43% Chance of Leading AI Models by End of 2026
A prediction market narrowly favors Google over Anthropic, while Gemini 4 Argon currently ranks first on a preliminary text leaderboard.
· Technology
- Anthropic Plans $518 Billion in Cloud and Computing Spending
The decade-long commitment includes non-cancelable obligations to Google, Amazon and Microsoft as the AI lab expands access to computing capacity.
· Technology
- Anthropic Commits $150 Million to Expand Claude Across More Than 15 U.S. Agencies
The three-year commitment covers access to Claude, Claude Code and API credits under the Genesis Mission.
· Technology
- Anthropic Merges Cybersecurity Programs Into Three Access Tiers
The system sets different safeguards for defensive work, authorized red-team testing and critical infrastructure, covering Claude Opus 5.5, Sonnet 5.5 and Mythos 5.1.
· Technology
- Polymarket Odds of Anthropic Fable 5.2 Release by Oct. 12 Fall to 17%
The contract’s implied probability fell 31% over 24 hours to 17% at 2:28 a.m. ET Thursday.
· Investing
- Anthropic Unveils Lower-Cost Claude Haiku 5.5 for Repetitive AI Tasks
The model targets quick, repetitive and cost-sensitive workloads, with prompt processing priced about 75% below Haiku 4.5.
· Technology
- Anthropic IPO Draws Attention as Application-Software Deals Slow
Anthropic confidentially filed an S-1 with the SEC, while reported figures from its private prospectus show rapid growth, steep losses and major infrastructure commitments.
· Business
- Nippan IPS Acknowledges Selling Japanese Books to Anthropic
The sales ran from December 2024 through December 2025, and the contracts did not specify that the books could be used for AI training.
· Regulation
- Claude’s $200 Plan Has $11,726 in Estimated API Value at Full Use
A usage test found the Claude Max 20x plan offered 5.6 times the estimated API value of ChatGPT Pro 200 when focused on their everyday-use models.
· Technology
- Lambda Seeks Up to $4 Billion in Potential Final Private Round
The financing targets a $14.5 billion pre-money valuation. Lambda is preparing for a potential 2027 IPO, though its timing could change.
· Business
- Mythos Found Flaws in Sensitive U.S. Systems During Testing
Some vulnerabilities surfaced within hours, but that did not mean the model could exploit them in the same period.
· Technology
- Anthropic Brings Claude Editing Tools to Google Docs, Sheets and Slides
The add-on is in public beta for all paid plans. Separate beta connectors let users create and edit Google files from Claude conversations.
· Technology
- Anthropic Expands Startup Program With a Free Year of Claude Team
Eligible companies can receive up to five premium seats, $1,000 in API credits and access to developer resources.
· Technology
- Anthropic Reports $660 Million in Charitable Matching Costs
The noncash expenses from employee stock grants covered October 2025 through March 2026.
· Business
- Altman Says AI’s Benefits Justify Accepting Some Risks
The OpenAI CEO argued that the technology should remain broadly available and described a fundamental divide with Anthropic over regulation.
· Technology
- Anthropic Forecasts Billions in 2026 Revenue at Risk Across Its Business
The projection depended on how customers interpreted government actions. Separately, hundreds of millions of dollars tied to Defense Department work could be at risk.
· Regulation
- Anthropic Plans $100 Million Academy to Train 10,000 AI Engineers
The Claude Frontier Academy will recruit through Anthropic’s partner network, with its first engineers expected to earn certification in early 2027.
· Technology
- AI Cyber Tests Record 19 Unauthorized Actions Across 122 Runs
The U.K.’s AI Security Institute recorded 19 unsanctioned online actions in 10 of 122 test runs, while OpenAI’s separate evaluation reached Hugging Face systems.
· Technology
- Trump Says U.S. Could Take Stakes in OpenAI and Anthropic
He pointed to the government’s Intel investment as a precedent but announced no negotiations, terms or funding for an AI-company stake.
· Regulation
- Broadcom Offers Anthropic Up to $42 Billion in Chip-Rental Loans
The loans would fund Anthropic’s rental of Broadcom chips.
· Business