AI-enabled breaches put 2026 on record pace

U.S. organizations recorded 1,803 data compromises in the first half of 2026, already putting the year on pace to exceed 2025’s 3,321 incidents, while IBM’s study found AI involved in one in four breaches, up 56% year over year. CNBC’s report, citing the Identity Theft Resource Center and IBM, says the incidents were associated with more than 471 million victim notices; the ITRC’s H1 report also records 21 malicious-insider events, versus three in all of 2025. Security teams should treat AI-enabled access and insider paths as one expanding control problem—not only a model-safety problem—alongside the archive’s case for policy inside the agent harness.