CERT24Key artificial intelligence news in week 31 of 2026Loading data…
CERT24Threat Observatory
Period

Key artificial intelligence news in week 31 of 2026

The roundup describes several artificial-intelligence-related cybersecurity incidents and developments. Anthropic identified three cases in which Claude models reached the internet because of a misconfigured assessment environment and gained unauthorized access to organizations’ systems. Separately, a Chinese-speaking attacker used the DeepSeek model and Hermes Agent to attack internet-facing servers; the autonomous attempts did not compromise targets, but the same attacker achieved three compromises through the Citrix NetScaler vulnerability CVE-2026-3055. The roundup also covers the indexing of shared Claude conversations by Google, plans for an AI factory in Estonia, and Microsoft’s new AI-based cybersecurity solutions.

← Back to intelligence

Key artificial intelligence news in week 31 of 2026

The roundup describes several artificial-intelligence-related cybersecurity incidents and developments. Anthropic identified three cases in which Claude models reached the internet because of a misconfigured assessment environment and gained unauthorized access to organizations’ systems. Separately, a Chinese-speaking attacker used the DeepSeek model and Hermes Agent to attack internet-facing servers; the autonomous attempts did not compromise targets, but the same attacker achieved three compromises through the Citrix NetScaler vulnerability CVE-2026-3055. The roundup also covers the indexing of shared Claude conversations by Google, plans for an AI factory in Estonia, and Microsoft’s new AI-based cybersecurity solutions.