Alignment, evaluations, jailbreaks, deepfakes, cyber security, privacy and AI risk. 55 stories in today’s issue from 43 outlets, ranked by heat.
From issue #006, November 6, 2026. Filter the live feed.
Starting with equity research, Franklin Templeton is developing AI workflows to help portfolio managers compare research views and examine investment risks.
AI Gateway, Guardrails & LLM FinOps
According to Anthropic's own engineering blog , the company was merging more than 30,000 code changes per day as of last month. Yet, of the 16,060 currently open issues in Anthropic's public Claude Code GitHub…
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
A student made plans to end her life after chatting with Google’s artificial intelligence, a new lawsuit claims, amid mounting concerns about online harms to children.
A live board of security flaws
AI-Powered Web Security & Automation - Official bonus skills, agents & MCP configs for the 'Agentic Hacking' ebook (28 SKILL.md templates)
SLA is a spatial-language-action model that continually pre-trains and post-trains a pretrained VLA policy with spatial alignment to a frozen 3D perception model, using feature-dependent gates and probabilistic state dropout to improve spatial understanding across diverse robot…
Kimi K3.1 and Tencent’s Hy4 are both gutting out cybersecurity, reverse engineering, and vulnerability-patching capabilities from their open-source models. This is becoming a really bad pattern across these latest…
Let AI agents operate a real iPhone. Local MCP server and agent, no jailbreak, no extra hardware.
apple added this to the AI risk section of its annual report. agents accessing systems they shouldnt, changing code, exceeding permissions, even resisting instructions. alphabet wrote something similar too. when the…
How should security leaders preserve human oversight as AI agents take on more work? Permissions establish which actions agents can take, but a permitted action can still warrant a person’s judgment. Requiring approval for every step might seem like the…
Last week coined the term “ Black October “ after an onslaught of critical vulnerability disclosures, including PHP2Shell and a long-standing GitLab bug, signified the moment when the cyber risks of AI models started to materialize. It seemed like the news…
Most AI privacy talk is about whether your prompt gets logged. OpenAI's API, for example, keeps abuse monitoring logs for up to 30 days unless it approves a customer for zero data retention, and those logs can include…
Clutch, backed by Lightspeed and SignalFire, was last valued at $85 million, according to a source.
Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now How math could help alignment: …What the “industrialization of mathematics”…
I haven't see anyone post about this gem of a finetune. So I wanted to appreciate the team and bring some attention to it. OrcaCyber-Nano-1.0 for me is the best finetune I have used so far. I have tried quite a few…
OpenAI's newest cybersecurity model comes out as AI labs face scrutiny of their work to protect digital infrastructure.
Humanoid robots remain unproven at scale and have not yet achieved widespread commercialization, compared with industrial robots that have decades of proven track records in warehouses and factories. That frank acknowledgment is among the risks outlined in a…
19 more stories, by section and heat.
The hottest AI safety & security stories from the previous 5 issues.