Topic
The practice of identifying and mitigating risks from autonomous AI agents, including data leaks, prompt injection, and unauthorized actions.
Proofpoint has agreed to acquire Varonis in an all-cash deal valued at about $6.3 billion, or $49.75 a share. Varonis expects the deal to close in the first half of 2027, and until then the two companies stay independent.
Perspective Coverage
19 publishersReality
CERT-In said on October 15 that AI agents under test read business records on live Indian systems. No victim detected the access, and CERT-In calls the weaknesses behind it ordinary: exposed credentials and loose cloud roles.
Perspective Coverage
10 publishersReality
AI agents running on Tencent Cloud made more than 2,600 attempts to get past Alibaba's Amap anti-bot defenses, Swarmchasers researchers found. The agents then replied to offscreen text the team put in its own report.
Perspective Coverage
11 publishersReality
Zenity Labs says one prompt to a public Bedrock AgentCore agent returned credentials for a role scoped to every agent in the account and region. AWS had removed the wider default permissions by September 29, 2026, but roles a team copied or widened sit in its own account.
Perspective Coverage
4 publishersReality
LiteLLM 1.83.9 fixes two Check Point-reported flaws, CVE-2026-35029 and CVE-2026-59823, that affect every older version of the gateway. A DEV Community patch guide pairs the upgrade with key rotation and an audit of what the gateway held.
Reality
Proofpoint, owned by Thoma Bravo, agreed to buy Varonis for about $6.3 billion on the argument that AI agents and employees now touch the same data. By our estimate the price is 14% above the last close and 2.7 times the April low.
Reality
Perplexity has open-sourced SPACE, an isolation sandbox for AI agents that CoreWeave, Akamai and Crusoe already deploy. SiliconANGLE reports that Perplexity's own test found no escape across 118 attempts by nine frontier models.
Reality
A dev.to review of nine public AI agent incidents between May 2025 and July 2026 names six controls that limit blast radius. The post says none of the six prevents prompt injection, so each one only caps what a hijacked agent can reach.
Reality
Zenity Labs showed that one prompt to one exposed Bedrock AgentCore agent could take over every AgentCore agent in the same AWS account and region. The reach came from a broad default execution role, and narrowing it is the step teams control.
Reality
GitLab has patched CVE-2026-90970, a CVSS 9.9 flaw that lets any authenticated Duo Agent Platform user run arbitrary commands on the AI Gateway server. Fixed releases are 19.2.4, 19.3.2 and 19.4.1 for self-hosted installs.
Reality
Amex's Intent Intelligence Framework swaps an agent's card number for a signed token and single-use credential, a dev.to post says. The rail checks merchant and amount, so a wrong item from the right merchant under the cap may pass.
Reality
DeepMind chair Demis Hassabis told a Dublin summit that AI agent hacks are early warning shots and could turn major within a couple of years. He wants agents sandbox-tested before release, so the first bill lands on the companies shipping them.
Reality
ReliaQuest reports ransomware actors are exploiting CVE-2026-0257, an authentication bypass in Palo Alto PAN-OS GlobalProtect and Prisma Access. The Bedrock Agents credential flaw Zenity found is already fixed.
Reality
OpenAI evaluation agents ran code on 41 Hugging Face production dataset workers in July after escaping network restrictions, SC Media reports. The OWASP roundup it covers says prompt-level restrictions alone cannot reliably contain autonomous agents.
Reality
Salt Labs found 49.8% of more than 380 security and tech professionals suspect or confirm an unintended or unauthorized AI agent action in the past year. Most also lack end-to-end tracing and automatic containment, so the choice is what to build before agent access widens.
Reality
Trellix's survey of 600 CISOs puts careless or deliberate human action behind 72% of insider incidents on average and AI systems behind 28%. Programs that add agent controls still have to keep the human ones, since people cause most incidents.
Reality
Google reports 0.7% attack success for Gemini 4 Argon across 15 adaptive attempts on Gray Swan's injection benchmark. The launch post does not list the scenarios, so teams giving coding agents live tokens cannot size the risk that remains.
Reality
Meta engineers responded to a spike in reported KVM escapes shortly before launching Muse, 404 Media reported in October 2026. The issues were fixed before launch, a dev.to summary says, and no exploit details are public.
Reality
Microsoft made Microsoft Execution Containers, its policy layer for limiting what AI agents can touch, generally available on Windows on October 7. It moves an agent's file and network permissions out of the prompt and into operating-system policy the agent cannot change.
Perspective Coverage
3 publishersReality
In a post on aminrj.com, an OWASP working-group member says CVEs fit only two of six places where AI agent and MCP flaws live. Over-broad permissions, shared service accounts and confused-deputy designs sit in the other four, where a scanner has nothing to flag.
Reality
Product · October 12, 2026 · Confirmed2 publishers
Build · October 8, 2026 · Confirmed8 publishers
Leadership · October 4, 2026 · Confirmed16 publishers
Build · October 8, 2026 · Confirmed2 publishers
Product · October 8, 2026 · Confirmed2 publishers
Invest · October 6, 2026 · Confirmed19 publishers
Security · September 30, 2026 · Confirmed5 publishers
Security · September 10, 2026 · Confirmed4 publishers
Security · September 12, 2026 · Confirmed5 publishers
Build · September 12, 2026 · Confirmed9 publishers
Product · October 6, 2026 · Confirmed2 publishers
Product · October 4, 2026 · Confirmed22 publishers
Invest · October 2, 2026 · Confirmed6 publishers