Anthropic launches programs to defend critical infrastructure and open-source software
publishers 72articles 107first reported 24 Sep, 16:43 UTCdeveloping since 24 Sep · 38 editions
Anthropic has launched a cybersecurity initiative focused initially on critical infrastructure and open-source software.
The company’s new Critical Infrastructure Defense Program will provide frontier AI models, on-site engineers and threat research to defenders responsible for operational technology behind power grids, water systems, factories and transportation networks, as well as government systems. Anthropic said such equipment often cannot be taken offline for upgrades, leaving known vulnerabilities unresolved because patches can be risky to deploy.
Anthropic also introduced OSS Scanner, a free, opt-in service offering open-source projects regular vulnerability scans using its strongest models, including Claude Mythos. The service is intended to identify security flaws and propose patches in widely used open-source code.
According to The Verge, OSS Scanner’s reports will be generated entirely by AI, without human review or triage. Anthropic acknowledged that this approach allows faster and more frequent scans but could produce incorrect or invalid reports. Some open-source maintainers have already struggled with an influx of AI-generated vulnerability submissions.
The two programs form the initial phase of the Anthropic Cyber Mission, a broader commitment to providing tools, research, engineering support and funding to security defenders. Anthropic said the effort builds on its Project Glasswing work and an expanded Cyber Verification Program, after concluding that finding vulnerabilities had become easier while verifying, prioritizing and fixing them remained difficult.
HOW THIS STORY WAS MADE
Written from 8 articles, 5 with the publisher's own text; 7 independent newsrooms once syndicated copies count as one; this version written 210 h after the record first saw the story; the editor kept it as written.
- Publishers
- 72
- Source articles
- 107
- Given to the writer
- 8, 5 with the article's own text
- Independent newsrooms
- 7
- This version written
- 210 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
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 →