AI & software security
-
AI agents used in live retail attacks
Gambit Security says a financially motivated operator used three open-source AI harnesses to automate vulnerability discovery, exploitation, and campaign orchestration against online retailers.
-
Mandiant warns attackers are targeting build pipelines
Mandiant says recent campaigns increasingly target developer tools, security scanners, CI/CD systems, and trusted software-delivery mechanisms rather than finished applications alone.
-
Gartner forecasts agent abuse in enterprise breaches
Gartner predicts ungoverned AI-agent abuse will drive 25% of enterprise breaches by 2028 as agent identity, authority, and oversight become security-control problems.
-
Omada buys EmpowerID for agent identity controls
Danish identity-governance vendor Omada has acquired EmpowerID, adding runtime controls for AI agents to a platform traditionally focused on human and non-human identities.
-
Dutch agencies warn AI is accelerating cyber threats
Dutch intelligence, cyber, police, and government bodies have jointly warned that AI is increasing the speed and scale of cyberattacks, placing greater responsibility on organisational leadership.
-
AI incident exercises remain uncommon
ISACA research finds only 8% of surveyed organisations regularly conduct AI-specific incident-response exercises, despite AI becoming increasingly embedded in day-to-day security operations.
-
Cyera adds $400m for AI security
Goldman Sachs Alternatives has invested another $400 million in Cyera as the data-security company expands into AI-agent oversight and non-human identity.
-
KDDI launches agentic MDR across EMEA
KDDI has launched a managed detection and response service for the US and EMEA using Exaforce technology to automate parts of security triage, investigation, and response.
-
Proofpoint joins AI and data controls
Proofpoint has announced a system intended to connect AI behaviour, data access, identity, and business intent, with key capabilities expected to become available by the end of 2026.
-
Darktrace expands controls for enterprise AI
Darktrace has made SECURE AI generally available, extending its behavioural monitoring into shadow AI, prompts, AI identities, and enterprise policy enforcement.