One shared ChatGPT link took over the signed-in account
A public ChatGPT share ran attacker JS as the signed-in origin. Parameter's write-up, OpenAI's fix, and a local rebuild that shows which check stops it.
Tags
32 posts tagged
A public ChatGPT share ran attacker JS as the signed-in origin. Parameter's write-up, OpenAI's fix, and a local rebuild that shows which check stops it.
MathLive 0.111.2 flipped originValidator from none to same-origin. One npm pack + grep shows it.
A tampered Claude Code extension sat on Cursor's marketplace for 11 hours on Oct 29. Which versions are real, what it planted, and a checker you can run.
After each reply, Anthropic adds one hidden turn and Claude returns a JSON safety flag in ~10 tokens, under 2% of compute. The KV cache makes it cheap.
Anthropic's Safeguards team published how Claude monitors its own conversations for misuse. After each reply, the system appends a short hidden instruction aski...
via Xssh-keygen -lf reads a whole authorized_keys file, handles option prefixes and prints one SHA256 fingerprint per key. Match bad keys on that.
Jules goes read-only after Nov 11, 2026 and shuts down Dec 31. Google's export runs through a CLI agent like Claude Code. Keep agent context in the repo.
Google is winding down Jules, its asynchronous coding agent. You can keep working in sessions until November 11, 2026, after which Jules turns read-only, and it...
Node 20.20.2's bundled roots still trust the compromised eMudhra CA. --use-openssl-ca helps until code passes ca: tls.rootCertificates.
Inside a bubblewrap sandbox started with --disable-userns, unshare fails with ENOSPC, 'No space left on device'. That error means the block works.
A KV cache pool smaller than the context window made Cloudflare's GLM-5.3 return other users' data in 155 responses. How it happened and the SGLang fix.
Cross-request KV cache exposure in the GLM-5.3 Workers AI model, and our remediation with SGLang (Kevin Flansburg and Michelle Chen, Cloudflare, October 26, 202...
via Hacker NewsLong-running AI agents lock onto one plan. Four October 2026 studies point to one fix: forget and restart. What it costs and what to hand off.
process.versions.brotli reports the same Brotli version before and after Node's CVE-2026-111497 fix. Check zlib.constants for the -17 error code instead.
"Our red-teaming agents found vulnerabilities in almost every one of these services that gave them limited forms of internet access."
— OpenAI
Kimi K3.1 Pro lists at half Opus 5.5's price, but cache hits narrow the gap. Session cost math, Moonshot's benchmarks, and the cyber data it removed.
The EU AI Act for developers: risk tiers, provider vs deployer, Article 50 transparency, GPAI rules and the AI Omnibus dates, with a Python triage script.
A system prompt is text placed before the conversation that sets a model's role and rules. See the raw chat template, its token cost and its limits.
An AI agent is a model that calls tools in a loop until a goal is met. A runnable Python loop, what each step costs in tokens, and where agents break.
Agent skills are folders with a SKILL.md that agents load in stages. The spec, a runnable demo, and what each stage costs in tokens.
MCP is an open protocol that lets AI apps find and call tools over JSON-RPC. See the four messages a host sends, with a runnable Python server and client.
Code execution with MCP lets an agent call tools from code, so definitions and big results stay out of context. How it works, with a runnable token count.
Agent sandboxing limits which files an AI agent can touch and which hosts it can reach, enforced by the OS. How Claude Code and Codex do it, with a Python demo.
ChatGPT Atlas puts an agent inside your browser. How prompt injection hides in page text, why filtering it fails, and where the real check has to live.
The lethal trifecta: an AI agent with private data, untrusted content and a way to send data out can be made to leak it. How to spot it and break the chain.
AI guardrails are checks around an LLM call that block, rewrite or route bad inputs, outputs and tool calls. Rail types, limits and a Python pipeline.
On 22 October 2024 Anthropic put computer use in public beta. Claude scored 14.9% on OSWorld screenshots. Build around the prompt-injection warning.
The xz backdoor (CVE-2024-3094) hid in xz 5.6.0 and 5.6.1 release tarballs to hook sshd. How it worked, how it was caught, and a tarball diff in Python.
AI red teaming means attacking your own model or LLM app to find harms and security holes before users do. Methods, tools and a runnable Python harness.
AI jailbreaking tricks a model into ignoring its safety training. Why it works, how it differs from prompt injection, and which defenses hold, in Python.
Prompt injection is when untrusted text in an LLM's context overrides your instructions. Direct vs indirect attacks, why filters fail, and designs that hold.
An SBOM (software bill of materials) lists every component in your software and how it got there. The NTIA fields, SPDX vs CycloneDX, and a Python parser.
Passkeys replace passwords with a key pair: the private key stays with you, the site keeps only a public key. How WebAuthn works and why it stops phishing.
eBPF lets you run small, verified programs inside the Linux kernel without writing a kernel module. How the verifier, maps and hooks work, with bpftrace.
Log4Shell (CVE-2021-44228) let attackers run code by getting one string into a Log4j log message. How JNDI lookups caused it, with a safe Python demo.