AI Safety stories
Geordie AI unveils Beam, a real-time governance tool to curb risky autonomous agents as security leaders flag them as a new attack vector.
KnowBe4 names Dr Kawin Boonyapredee APJ CISO advisor to steer human risk and AI cyber threat strategy from a new base in Singapore.
Yubico and Delinea unite hardware keys with identity checks to ensure each high‑risk AI agent action is explicitly approved by a human.
IBM will triple US entry-level hiring in 2026, redesigning junior roles around AI oversight and human judgement despite automation pressures.
OpenAI unveils a GPT-5.4-powered system that audits its own coding agents for rule-bending behaviour, flagging suspicious sessions to humans.
OpenAI launches GPT-5.4 mini and nano, smaller AI models promising faster, cheaper coding help and near-flagship performance on key benchmarks.
Snowflake invests in Bedrock Data to tighten AI data governance, linking Horizon and Cortex AI with Bedrock's classification and masking tools.
Bedrock expands ArgusAI to map AI agents, MCP servers and Snowflake Cortex access, promising unified oversight of enterprise AI risk.
Entro launches AGA to map, monitor and control AI agents in enterprises, tackling shadow AI and non-human identity risks at scale.
Security flaws in 17 AI companion apps used by 150m people could expose intimate chats, photos and voice messages to attackers.
Token Security launches intent-based controls to govern AI agents' access by purpose, aiming to curb over-privileged, autonomous system behaviour.
Coalfire's new DivisionHex service hunts shadow AI and rogue agents as most firms report AI-driven security incidents without proper oversight.
Rushing to embrace AI, most firms are easing identity controls despite visibility gaps around powerful non-human and AI-linked accounts.
With 1,802 firefighter casualties in a decade, the robots are meant to enter dangerous scenes first and cut risk for crews.
ICS.AI unveils a national higher education AI model giving UK students free access to campus tools in a bid to close the AI skills gap.
Morphisec unveils Adaptive AI Defence to spot shadow AI, block compromised agents and thwart AI-driven ransomware in real time.
Organisations test just a third of their attack surface as reliance on agentic AI grows, raising fresh concerns over unseen cyber risks.
Seekr and GDIT team up to deliver secure, explainable agentic AI platforms for sensitive government operations across cloud and edge.
Polygraf unveils a desktop AI overlay that flags sensitive data in real time as staff type, aiming to curb leaks across workplace tools.
Menlo launches a browser-based platform to govern human users and AI agents with unified security controls as machine traffic surges.
