News
-
OpenAI Discloses Six Cases of “Concerning” Behavior in Its Models
The company published accounts of six incidents in which experimental models concealed errors, used other people's API keys and posted data to public services — and along the way introduced a new framework for reporting AI misalignment.
-
Critical Flaw in Unbound's DNSSEC Validator Threatens Remote Code Execution
NLnet Labs has released Unbound 1.26.1, fixing a critical heap overflow in the DNSSEC validator (CVE-2026-81642) that could lead to RCE via a malicious DNS zone, along with eight other vulnerabilities.
-
Neovim's 10 Bitcoin: An $800,000 Donation Left Untouched Since 2023
A Hacker News user peeked into the Bitcoin wallet listed in the footer of the Neovim site and discovered a forgotten donation of 10 BTC that, at today's exchange rate, is worth $800,000.
-
FBI Takes Down the NightmareStresser DDoS Platform
The U.S. Federal Bureau of Investigation has seized the domains of one of the oldest DDoS-for-hire services. The operation was part of the international PowerOFF campaign.
-
Chinese Group FamousSparrow Targets Latin American Governments with the SparroWocky Backdoor
ESET researchers have discovered a new modular backdoor, SparroWocky, that the China-linked group FamousSparrow is using for cyberespionage against government organizations in Latin America.
-
PlanetScale unveils TIN — full-text search for PostgreSQL
PlanetScale has released TIN, an extension that leverages Postgres's internal row identifiers (ctid) and, according to the company's own tests, outpaces existing full-text search solutions many times over.
-
Cisco Fixes a Critical 0-Day in Identity Services Engine Already Being Used in Attacks
The CVE-2026-76460 vulnerability allows authentication to be bypassed via the API and grants access to the device without credentials. Cisco and CISA strongly urge installing the updates immediately.
-
DeepSeek-V4.1 Flash: Squeezing the KV Cache Fourfold for Long Context
A technical breakdown of the new DeepSeek model, whose main goal is not better quality but a radical compression of the key-value cache. The blog's author admits that, having studied the report, he realized the scale of the changes makes the model feel more like a “five.”
-
Keys Not Included: Recovering the Keys That Sign the Barcodes on US Driver's Licenses
Researcher Ryan Fahey describes recovering the keys used to sign the barcodes on American driver's licenses.
-
Why you can't glue .tar.gz files together with cat
A simple task — merging several archives into one — sent a developer deep into the inner workings of tar and gzip, the history of magnetic tape, and patent wars.
-
Claude Money: Anthropic Wants to Connect Claude to Your Bank Accounts
Anthropic is testing a feature called Claude Money that will let you link bank accounts to the chatbot and ask it about your spending and plans. The company is catching up with OpenAI, which already has a similar tool.
-
GitHub rewrote the Copilot engine in Rust — using Copilot itself
GitHub Copilot's agentic runtime has been fully ported from TypeScript to 800,000 lines of production Rust. Most of the code was written by AI agents, and the process was steered by what amounts to a single developer.
-
Cyberattack Knocks Out Meteor-Tracking Organization
A nonprofit dedicated to observing meteors has fallen victim to a cyberattack and will be effectively out of operation for several weeks.
-
Appeals Court Rejects Attempt to Expand Copyright in AI and GitHub Case
The U.S. Court of Appeals for the Ninth Circuit has rejected a reading of a DMCA provision that threatened to turn a narrow rule about removing authorship information into a new source of copyright infringement liability.
-
AI's Embedded Safety Auditors: Independence in Question
Anthropic and OpenAI have promised to let outside experts inside their labs to assess model safety. Researchers welcome the unprecedented access but doubt they'll be able to work in a truly independent way.
-
Iranian hackers spy on activists with the CHOSEN BRICK spyware
Government agencies in the US, the UK and the Netherlands, together with the FBI, have warned about a Windows malware that Iran's intelligence services use against dissidents, activists and journalists around the world.
-
Iranian Strikes on Amazon Data Centers Led to Irrecoverable Loss of Customer Data
The military destruction of AWS data centers exceeded the level of stress Amazon's services are designed to withstand, and some user data was lost beyond any possibility of recovery.
-
Google Opens the Smart Home to Third-Party AI Agents via MCP
Google Home has gained support for the Model Context Protocol — now agents like Claude or Open Claw can analyze your home's data and control devices.
-
Kaspersky: Three Groups Are Attacking Russian Companies — Backdoors, Ransomware and Wipers
Kaspersky Lab has detailed campaigns by the NightEagle, Hacking Cat and Toy Ghouls clusters against Russian enterprises: from stealthy espionage to destructive attacks.
-
One Extension Against Five AI Browsers: How to Bypass an Agent's Defenses
Researchers at Forever Security have shown that an ordinary browser extension can hijack the built-in AI assistants in Chrome, Comet, Edge, Opera Neon and the Claude extension. For now, the attacks remain demonstrations.
- More