AI safety concerns grow as firms drop pledges and security alerts flood teams

A Check Point report finds that AI‑driven attacks are overwhelming security teams with a surge of warnings, yet only about 7.8 % of those alerts are truly critical. The study highlights a widening “exposure gap” where organisations struggle to prioritize and remediate high‑risk vulnerabilities, with sectors such as utilities and government seeing most critical exposures from software flaws, while health‑care and finance face more data‑leak risks.

Separately, the Future of Life Institute’s AI Safety Index shows major AI developers—including Anthropic, OpenAI, Google DeepMind and Meta—have weakened or withdrawn prior safety commitments. All nine companies scored low, and experts at a UN AI conference warned that the industry is “running toward a cliff,” urging global safety standards and tighter regulation.