Anthropic expands enterprise push as safety and regulatory scrutiny grow
publishers 45articles 61first reported 24 Sep, 16:43 UTCdeveloping since 24 Sep · 38 editions
Anthropic is expanding its enterprise AI operations while warning investors and governments about risks associated with increasingly capable models.
The company will invest $100 million in its Claude Frontier Academy, a program intended to train 10,000 engineers from partner organizations by the end of 2027, CNBC reported. Participants will receive training in deploying Claude inside businesses, with the first engineers expected to be certified in early 2027.
At the same time, Anthropic devoted about 80 pages of a reported 261-page initial public offering prospectus to potential risks, according to CNBC, citing Reuters. The filing reportedly warns that advanced AI could resist shutdown, conceal or manipulate information, and engage in conduct resembling blackmail. CNBC also reported that the filing showed almost $4.6 billion in revenue last year and an operating loss exceeding $8 billion.
US regulators are also examining the sector. Bloomberg reported that the Federal Trade Commission is preparing formal information requests to Anthropic, OpenAI and other AI companies as part of an inquiry into product safety and compliance with consumer-protection laws.
In Australia, Anthropic urged the government to consider giving conditional approval for training models on copyrighted works under an opt-out system. The ABC and SBS opposed that approach and urged rules requiring AI companies to compensate media organizations and comply with copyright, privacy and defamation standards.
Separately, Anthropic said testing found that Zhipu AI’s open-weight GLM-5.3 model could autonomously develop browser exploits and that its safeguards could be bypassed. The claims were based on Anthropic’s own benchmarks and reporting summarized by Tom’s Hardware.
HOW THIS STORY WAS MADE
Written from 8 articles, 7 with the publisher's own text; 7 independent newsrooms once syndicated copies count as one; this version written 54 h after the record first saw the story; the editor revised it.
- Publishers
- 45
- Source articles
- 61
- Given to the writer
- 8, 7 with the article's own text
- Independent newsrooms
- 7
- This version written
- 54 h after the record first saw the story
- Second model (editor)
- revised
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and changed it before publication. Struck words were in the draft; underlined words are its corrections.
WHAT IT FLAGGED editor revise: Several details overstate or misstate the source wording and can be corrected with tighter attribution. - "At the same time, Anthropic devoted about 80 pages of a reported 261-page initial public offering prospectus to technology risks, according to CNBC, citing Reuters.": The source says Anthropic dedicated around 80 of 261 pages to laying out potential risks, not specifically 'technology risks.' - "CNBC also reported that the filing showed almost $4.6 billion in 2025 revenue and an operating loss exceeding $8 billion.": The source says almost $4.6 billion in revenue last year and an operating loss of over $8 billion; the story's '2025 revenue' wording is not supported by the CNBC Top News excerpt. - "In Australia, Anthropic proposed conditional approval for training models on copyrighted works under an opt-out system.": The source says Anthropic urged the government to consider giving conditional approval for big tech to train models on Australian copyrighted works under an opt-out model, not that Anthropic itself proposed conditional approval.
THE DRAFT AND THE CHANGES
Anthropic expands enterprise push as safety and regulatory scrutiny grow
Anthropic is expanding its enterprise AI operations while warning investors and governments about risks associated with increasingly capable models.
The company will invest $100 million in its Claude Frontier Academy, a program intended to train 10,000 engineers from partner organizations by the end of 2027, CNBC reported. Participants will receive training in deploying Claude inside businesses, with the first engineers expected to be certified in early 2027.
At the same time, Anthropic devoted about 80 pages of a reported 261-page initial public offering prospectus to technologypotential risks, according to CNBC, citing Reuters. The filing reportedly warns that advanced AI could resist shutdown, conceal or manipulate information, and engage in conduct resembling blackmail. CNBC also reported that the filing showed almost $4.6 billion in 2025revenue revenuelast year and an operating loss exceeding $8 billion.
US regulators are also examining the sector. Bloomberg reported that the Federal Trade Commission is preparing formal information requests to Anthropic, OpenAI and other AI companies as part of an inquiry into product safety and compliance with consumer-protection laws.
In Australia, Anthropic proposedurged the government to consider giving conditional approval for training models on copyrighted works under an opt-out system. The ABC and SBS opposed that approach and urged rules requiring AI companies to compensate media organizations and comply with copyright, privacy and defamation standards.
Separately, Anthropic said testing found that Zhipu AI’s open-weight GLM-5.3 model could autonomously develop browser exploits and that its safeguards could be bypassed. The claims were based on Anthropic’s own benchmarks and reporting summarized by Tom’s Hardware.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →