Skip to content
Sign Up for Our Newsletter
About
Close About
Open About
About the Institute for Security and Technology
Our Team
Board Of Directors
Careers
Contact Us
Featured Events
Cyber Policy Awards
Critical Effect DC
Projects
Close Projects
Open Projects
AI and NC3
Pioneering action-oriented efforts to explore how advanced AI capabilities will be integrated into nuclear command, control, and communications
AI Antitrust and National Security
Exploring how to more effectively account for national security considerations in AI antitrust cases while respecting precedent, scope, and the core principles of antitrust law
AI Risk Reduction Initiative
Assessing the emerging risks and opportunities of AI foundation models and developing risk reduction strategies
AI Chip Export Control Initiative
Safeguarding U.S. national competitiveness by closing critical compliance and enforcement gaps
AI Risk Barometer
Measuring national security professionals’ perceptions of AI futures through a technically-informed survey
CATALINK
Preventing the onset or escalation of conflict by building a resilient global communications system
Energy FIRST
Powering U.S. and allied security & prosperity through a resilient energy future
Ransomware Task Force (RTF)
Combating the ransomware threat with a cross-sector approach
Religious Voices and Responsible AI
Engaging religious communities on safe and beneficial AI
SL5 Task Force
Strengthening AI security through a multistakeholder approach
UnDisruptable27
Driving more resilient lifeline critical infrastructure for our communities
All Projects
» Explore all of IST's projects, past and current
Focus Areas
Future of Digital Security
Geopolitics of Technology
Innovation and Catastrophic Risk
Events
Insights
Contact
Search
Donate
Archive
scheming
Blog
Q&A: An AI Loss of Control “Warning Shot”
How might you lose control of a system as complex as a frontier AI model? In this month's newsletter, we speak to the authors of IST's newest report diving into the risk of AI Loss of Control – that is, models diverging from authorized constraints, to the extent that a human operator can no longer prevent or constrain undesired outcomes - and presenting a framework for mitigation.
AI
,
AI Loss of Control
,
artificial intelligence
,
deception
,
model drift
,
risk reduction
,
scheming
February 19, 2026
Blog
Something Mysterious Is Happening
In a blog reflecting on her co-authored report, AI Loss of Control Risks, Mariami Tkeshelashvili writes, "There is a difference between embracing mystery as a driver of discovery and being incurious about risks that are already becoming visible. Monitoring AI behaviours that we cannot fully explain is not pessimism—it is responsibility. It is what keeps us safe."
AI
,
artificial intelligence
,
deception
,
Indications and Warning
,
loss of control
,
manipulation
,
monitoring
,
risk reduction
,
scheming
,
strategic foresight
February 19, 2026
Report
AI Loss of Control Risk: Indications & Warning
Though technologists and policymakers alike are eager to address AI Loss of Control–a state in which an AI system diverges from authorized constraints–there are significant gaps in the ways stakeholders understand, anticipate, and perceive this risk. "AI Loss of Control Risk" proposes applying the Indications & Warning (I&W) methodology used by the intelligence community to monitor this risk.
AI
,
AI Loss of Control
,
artificial intelligence
,
deception
,
model drift
,
risk reduction
,
scheming
February 19, 2026
Search
Search
Home
About
CATALINK BRIEF
FAQ
Our Team
Why do we need crisis communications?
Activities
Events
Insights
Podcasts
Press
Get In Touch
