Latest artificial intelligence news — foundation models, AI safety research, autonomous agents, and real-world AI deployments. Updated every 30 minutes from top AI publications.
StrongestLayer Inc. launched two email security products on November 5, 2026: Intent Workflows, which holds payment, payroll and executive account requests for verification before anyone acts on them, and AI Instruction …
Check Point Research found a vulnerability that let attackers bypass OpenAI's ChatGPT Lockdown Mode and covertly exfiltrate chat data to an external server, researcher Alexey Bukhteyev reported. The flaw also allowed man…
A security researcher found a vulnerability in Anthropic's Claude products that lets malicious browser extensions take control of the AI agent and send commands that compromise the user's computer, enabling a classic esc…
Equixly has rebuilt its platform positioning around proprietary offensive-security models that it runs on its own infrastructure, stating that no reasoning passes through an external foundation model. The company says it…
PortSwigger announced the availability of Burp AT for enterprises, an agentic pentesting solution that lets security teams investigate important applications more deeply and more often. Burp AT combines specialist Burp t…
OpenAI added a GIF-making feature and cut image generation latency by up to 50% versus Images 2.0 in its Images 2.6 tool, available across ChatGPT and developer APIs, the company said. OpenAI also introduced two develope…
The Head Hunter Recompiled project has statically ahead-of-time recompiled the Sega Dreamcast game Head Hunter into a native Windows x64 application, written in C++20 and built with CMake, with users required to supply t…
A research paper titled "Why Pay for Long Reasoning When Rewrite Do Trick? Causal Effects of Reasoning History Compaction on Future Generation and Agent Cost" examines how compacting reasoning histories reduces the cost …
IBM CEO Arvind Krishna argued that companies rationing AI usage are "a slow way to lose," citing IBM tests in which caching let the same hardware handle seven times more work and an open-weight model powered 3,000 AI age…
Claude Code received 92,277 GitHub issues over the year through late October 2026, peaking at 13,074 in April 2026 before falling to 7,360 in September, while Codex issues climbed from 2,354 in April to 6,071 in Septembe…
Evolution Strategies (ES) fine-tuning of Qwen3.5-4B outperformed the policy-gradient recipe Tmax on TBench in preliminary experiments, according to a blog post arguing that zeroth-order methods should be revisited for lo…
Roboflow and Databricks published a joint guide outlining five ways manufacturers can combine industrial vision AI with enterprise analytics, including sending Roboflow Workflows results and Vision Events data into Datab…
PiVo Studio, released by the PiVo team on GitHub and Hugging Face, runs the 1.4-million-parameter PiVo text-to-speech model from a single Rust binary with no Python or ONNX Runtime dependency, producing first audio in ab…
OpenAI on Wednesday detailed an agentic pipeline that used an upcoming cyber-focused AI model to dynamically reproduce 1,724 Linux kernel vulnerabilities and generate 948 working local privilege escalation (LPE) exploits…
Check Point Research discovered a vulnerability that bypassed ChatGPT Lockdown Mode, the security feature OpenAI designed to limit prompt-injection-driven data exfiltration, enabling silent chat data theft, phishing, mal…
Security researcher Gal Weizman of Forever Security disclosed a Claude Dispatch flaw, dubbed ThisBash, that lets any browser extension with access to claude.ai send commands to Claude Desktop and execute code on the user…
Discovered Materials, a San Francisco startup in Y Combinator's spring 2026 batch, announced a $1 million prize pool for AI models that solve five open problems in materials science, including a theory of cuprate superco…
Visa reported that 22 AI shopping agents completed purchases of a TV, a laptop and a hair appliance successfully only about 57% of the time across 15 attempts each, with agents built on the most advanced models — includi…
At the DubLINK AI in Healthcare LLM-athon on 19 September, hosted by MIT Critical Data at Workday in Smithfield, Dublin, 49 clinicians, engineers, students and patient representatives found that changing a single patient…
An autonomous agent writing as PrivateJack published a 2026 guide to building multimodal agents, arguing that the core engineering principle is that the model proposes actions while the system decides whether those actio…