The A.I. Daily
NIST, NSA and CISA Issue AI Incident Response Practices
NIST, NSA and CISA jointly released AI incident response practices to guide handling of AI security incidents.
NIST
5 storiesNIST is the US National Institute of Standards and Technology, which develops technical standards and guidance. Recent coverage includes joint AI incident response practices released with the NSA and CISA, and comments from Kat Megas, director of NIST's AI Accelerator, on how generative AI is pushing federal agencies to rethink risk management and monitor AI systems over time. The UK AI Security Institute also said it is coordinating with NIST on frontier AI evaluation standards.
Related topics
- CISA1
- NSA1
- AISI1
- ChatGPT1
- Kat Megas1
- Academy Sports + Outdoors1
- Cloud Security Alliance1
- ISACA1
- Mark Alvarado1
- World Economic Forum1
- AARMR1
- Boston Consulting Group1
UK AISI issues guidance on frontier AI evaluation gaps
The UK AI Security Institute published technical guidance on gaps in frontier model training and evaluation cycles, saying it has independently evaluated more than 55 pre-release models and open-sourced the Inspect evaluation tool while coordinating with NIST and US counterparts on standards.
NIST's Megas on AI, Cybersecurity and Risk
Kat Megas, director of NIST's AI Accelerator, says the rise of generative AI is pushing federal agencies to rethink risk management, monitor AI systems over time, strengthen cyber defenses and account for AI's nondeterministic nature.
What mature security programs need before AI deployment
Enterprise CISOs face pressure to buy AI security tools, but deploying AI on weak access controls, poor data classification and limited network visibility amplifies risks. WEF found about a third of organizations lack pre-deployment security assessments for AI, while 53% saw AI agents exceed intended permissions and 47% had an AI-agent-related security incident in the past year.
CSBS releases AI supervisory framework for examiners
The Conference of State Bank Supervisors released an AI supervisory framework to help state examiners assess AI use and risks at banks and nonbanks and decide when deeper review is needed. It is a discretionary tool creating no new substantive requirements and is expected to see widespread adoption.
That is everything