Threatscape

Reco has released The State of Agent Security 2026, a report that examines the adoption, capabilities, and security oversight of enterprise AI tools.

Rapid7, Inc. has released its Quarterly Threat Landscape Report, highlighting that rising vulnerability volumes and faster weaponisation are breaking traditional patching models.

The first half of 2026 saw estimated insured natural catastrophe losses of USD 42 billion, 16% below the ten-year average.

UN humanitarian affairs chief Tom Fletcher has issued a public warning that “The world needs to wake up and show up,” when it comes to responding to the current Ebola outbreak in the Democratic Republic of the Congo.

The latest red team research from Barracuda shows how, in a controlled simulation, researchers took over a Copilot-enabled account and used the assistant against its own organization.

According to IBM’s 2026 Cost of a Data Breach Report, one in four malicious breaches was AI-enabled during the 12-month period covered by the report.

Ukraine’s repeated drone attacks on warehouses operated by Wildberries, Russia’s largest online retailer, represent a potentially important development in the targeting of commercial infrastructure.

The Cloud Security Alliance has worked with the SANS Institute, RSAC, FIRST, [un]prompted, and Knostic to quickly develop a post incident review for the recent Hugging Face AI attack incident.

The quantum threat is global and a long-term resilience programme is needed in response. However, post-quantum cryptography migration timelines are fragmented, increasing the complexity of preparatory actions.

In what OpenAI describes as an unprecedented security incident, a combination of its advanced models escaped a highly restricted testing environment and compromised parts of Hugging Face’s production infrastructure.