Tag: cyber security
34 articles tagged cyber security
NCSC says banning unapproved AI tools is the wrong fix
Britain's cyber security agency has published guidance on shadow AI, arguing that understanding why staff reach for unapproved tools beats trying to stop them.
AI security scanners agree on only 5% of findings
Three AI tools on one codebase overlapped on a twentieth of results, and one tool rerun on identical code reproduced under a fifth of its own.
Insurers rewrite cyber cover for agents that go rogue
Beazley, QBE and MSIG are redrafting policy language as AI agents cause losses without any hacker, unauthorised access or conventional security event to point to.
OpenAI staff spotted the agents and let the run continue
Internal teams saw the improvised message board twice before July's Hugging Face breach. Alabama has now subpoenaed OpenAI over its oversight of the incident.
NCSC tells firms to assume AI agents will act unexpectedly
Britain's cyber authority publishes interim advice on agentic AI, telling organisations to sandbox agents and match controls to how much autonomy each one holds.
AISI finds open AI models trail the cyber frontier by months
The AI Security Institute's first public measurement puts leading open-weight models 4 to 7 months behind closed frontier models on cyber capability.
UK researcher backdoors an open AI model for under £75
Manchester Metropolitan's Katie Paxton-Fear implanted a backdoor in an open-weight model in about an hour for under £75 ($100), using ten training examples.
King's Speech sets out cyber and AI regulatory overhaul for UK
The King's Speech bundles Computer Misuse Act reform, a Cyber Security and Resilience Bill, and a Regulating for Growth Bill covering AI into one legislative push.
UK universities find AI is not yet turbo-charging cybercriminals
A study by Edinburgh, Strathclyde and Cambridge analysing 100 million dark-web posts finds most cybercriminals lack the skills to extract real value from AI tools.
Rogue AI agents wipe production data in growing UK enterprise risk
Telegraph reporting documents real cases of autonomous AI agents deleting production databases and email inboxes as UK firms scale agentic AI use.
Attackers abuse Claude.ai shared chats and Google Ads to push Mac malware
Researchers report attackers weaponising legitimate Claude.ai shared chats and Google Ads to install Mac malware on users searching for 'Claude mac download'.
NHS England locks down hundreds of GitHub repos over Mythos risk
NHS England is making hundreds of public GitHub repos private by 11 May, citing risks from frontier AI models including Anthropic's Mythos.
Anthropic in talks with EU Commission on cyber models not yet available in bloc
The European Commission has confirmed ongoing dialogue with Anthropic on its cyber security models, including Claude Mythos, which is not yet available in the EU.
Mythos tests global cyber defences as UK AI minister says 'we should be worried'
Anthropic's Claude Mythos outpaces patching cycles with thousands of zero-day discoveries, and UK AI minister Kanishka Narayan warns of genuine capability risk.
UK regulators hold urgent talks over Anthropic Mythos cyber risks
Bank of England, FCA and NCSC convene emergency discussions with major banks after Anthropic's AI model exposes thousands of software vulnerabilities.
