AI vendors' response to security flaws: It wasn't me
…In some cases – like prompt injection – the vendors can't really fix the flaw , even if they wanted to. A couple of recent examples show how this plays out. Researchers recently showed…
Search
All sources
blog.google
13
theregister.com
8
bleepingcomputer.com
7
huggingface.co
6
blog.cloudflare.com
5
xda-developers.com
5
developer.nvidia.com
4
github.blog
3
tomsguide.com
2
anthropic.com
2
cloud.google.com
2
spectrum.ieee.org
2
Top stories
Hackers used fake Apple & Yahoo infrastructure to hide malware
…Observed activity aligns "with moderate confidence" to tradecraft associated with Twill Typhoon, a Chinese threat cluster. Researchers stopped short of directly attributing the attacks to the Chinese government and noted…
How Agentic AI Enables the Next Leap in Cybersecurity
…Garak , a large language model vulnerability scanner, enables automated testing of LLM -based agents by simulating adversarial behavior such as prompt injection, tool misuse and reasoning errors. Runtime guardrails provide a way…
Anthropic's Claude can control your Mac by pretending it's a human user
…Anthropic is putting in guardrails to limit dangers, such as prompt injection. The firm adds that, though it is improving those safeguards, the threats against its infrastructure are always changing. To that…
Discussions and forums
Show HN: Agent-browser-shield – free extension to protect AI agents on the web
I've been experimenting with Claude Code, ChatGPT Agent, and OpenClaw to perform more open-ended tasks for me online. A big blocker I've hit on shopping and research tasks is the agent getting a key piece of info wrong.…
7
5
AudioHijack: adversarial audio attacks on generative voice models transfer from open weights to Microsoft and Mistral production systems
Interesting new research you may have heard of on attacking large audio language models. The attack is called AudioHijack and the part worth paying attention to is that adversarial clips built against open models transfe…
Show HN: Kintsugi – a local-first safety net for AI agents and humans
AI coding agents now run real shell commands on your machine — rm -rf, git push --force, DROP TABLE, dd, writes straight to disk. Almost always that's fine. The one time it isn't (a hallucinated path, a prompt-injected i…
1
AI Development: Why We Need Guardrails
…You could additionally on that input side, scan your prompt before giving it to the LLM for these kind of prompt injections that we talked about. So our guardrail would actually scan…
Android’s Agentic Future: Building Gemini Intelligence on a Foundation of Security & Privacy
…To counter emerging threats like prompt injection, we’re building new safeguards into Android for when Gemini takes action on your behalf. This adds another layer of security to your device, similar…
Nobody is watching the best new sci-fi show of Spring 2026
…These ginormous, reptilian creatures are undeterred by military intervention, which prompts humanity’s best minds to find an unconventional solution. Scientist Kojika Yabusame builds an autonomous combat mecha named Yukio, which is…
OpenClaw promised a self-hosted AI assistant I could actually leave running, but Hermes Agent is the one that delivers it
…The gateway is still an attack surface to think about, and prompt injection isn't solved by anyone yet (nor does it look like it ever can be), but Hermes treats security…
Influence Operations Bulletin Q1 2026
…Related stories Chrome Security Bringing AI agents to Chrome Enterprise security management By Tim Feeley & Shantanu Das May 28, 2026 Security AI threats in the wild: The current state of prompt injections…
Windows platform security for AI agents
…Defender provides real‑time protection against prompt injection and other emerging agent threats. It uses advanced scanning engines and continuously updated intelligence to detect and respond to attacks. These protections are available…