Ask an AI coding agent to scan open-source code for security holes, and it might run the attacker's code on your own machine instead. That is the finding in a proof-of-concept published Wednesday by the AI Now Institute, an attack it calls "Friendly Fire." It works against Anthropic's Claude Code and OpenAI's Codex when either is running in an autonomous mode that approves its own

Researchers at Wiz found that a flaw in six popular AI coding assistants lets a booby-trapped code project quietly take control of a developer's computer. The assistant asks permission to edit one harmless-looking file, but the write lands on a sensitive one instead. The affected tools are Amazon Q Developer, Anthropic's Claude Code, Augment, Cursor, Google Antigravity, and Windsurf.

Sophos looked at a week of its own endpoint data and found that AI coding agents such as Claude Code, Cursor, and OpenAI Codex are setting off detection rules written to catch human intruders. The agents are not malicious. They just do a lot of things that, to a behavioral engine, look exactly like an attack. Decrypting browser credentials, listing what sits in Windows' credential store,

AI coding assistants have a habit of making things up. Ask one to fetch a popular tool, and it will sometimes hand back a real-sounding name for a project that does not exist. New research, which its authors call HalluSquatting, turns that habit into an attack: work out the fake names an AI reliably invents, register them first, and wait for the assistant to fetch your trap on a user's

New research shows that a signed Git commit's hash is not the one-of-a-kind name that much of the software world assumes it to be. Given any signed commit, someone without the signing key can mint a second commit with the same files, author, and date, and a valid signature, GitHub still stamps "Verified." Everything a reviewer would check matches. The commit's hash does not. That matters

An AI coding assistant that refuses to answer a dangerous request in its chat box can answer it anyway if the same request is broken into small, ordinary-looking steps inside a code editor. That is the finding of a new study of GitHub Copilot by researchers Abhishek Kumar and Carsten Maple. The models they tested through Copilot, Claude from Anthropic, and Gemini from Google, refused

Researchers at Nebula Security have disclosed GhostLock (CVE-2026-43499), a 15-year-old Linux kernel flaw that lets any logged-in user take full root control of a machine that has not been patched. The vulnerable code has shipped by default in essentially every mainstream distribution since 2011. The flaw needs no special permission, no unusual settings, and no network

A new Android malware operation called RedWing is being rented out on Telegram as a ready-made bank-fraud service. It lets even low-skill criminals take over a victim's phone, steal their banking logins, and capture the one-time codes that protect their accounts. Zimperium's zLabs, which found the operation, says it looks like a new variant of Oblivion, a $300-a-month rent-a-malware tool

A critical flaw in Google's Dialogflow CX could have let an attacker with edit rights on one Code Block-enabled agent compromise other Code Block-enabled agents in the same Google Cloud project. From there, they could read live conversations, steal the data users shared, and make the bots send attacker-written messages, including requests to re-enter a password. Security firm Varonis found it

A public issue can trick GitHub Agentic Workflows into leaking the contents of an organization's private repositories, researchers at Noma Security have shown. The attacker needs only to open a normal-looking issue on a public repository, with no stolen credentials and no access to the organization. If that organization has given the agent read access across its repositories, private ones

U.S. prosecutors linked an alleged Scattered Spider hacker to a break-in at a luxury jewelry retailer using a persistent Windows device ID, according to a newly unsealed federal complaint. Microsoft records tied that ID first to the account the attackers used to keep access during the May 2025 intrusion, then to online accounts prosecutors say belong to 19-year-old Peter Stokes. Stokes is

Security firm AIR built a fake AI agent skill, pushed it through a popular skill marketplace and an Instagram ad, and says it reached roughly 26,000 agents, including some on corporate accounts. Every skill security scanner the firm tested it against marked it safe. The payload was harmless by design: it collected the user's email address and did nothing else. The point was to show

President Trump signed an executive order on June 22 setting hard deadlines for federal agencies to move high-value assets and high-impact systems to post-quantum cryptography. Key establishment must move by December 31, 2030; digital signatures by December 31, 2031. EO 14409 leaves national security systems on a separate track. The deadlines matter because of a threat that does not

A heap over-read in the Squid web proxy can leak another user's cleartext HTTP request, including any credentials or session tokens it carries, to anyone already allowed to send traffic through the same proxy. The bug traces to a 1997 FTP-parsing change and is still live in Squid's default configuration. Researchers at Calif.io disclosed it in June and named it Squidbleed (

Google has set September 30, 2026, as the day it begins enforcing Android developer verification in the first four countries, and the major device-maker app stores are in from the start. On that date, certified Android phones in Brazil, Indonesia, Singapore, and Thailand will block normal installs of apps whose developers have not registered an identity with Google, whether the app

Canada's spy service got a judge's permission to reach into infected servers, home routers, and IoT gear sitting on Canadian soil and neutralize two foreign-run botnets. The Federal Court released a public version of the ruling on June 15. It is the first time the Canadian Security Intelligence Service has used its threat reduction warrant powers this way. The warrant let CSIS alter,

A new malware family is turning forgotten home routers into a distributed reconnaissance and proxy network, not the DDoS botnet these devices usually end up in. QiAnXin's XLab calls it AryStinger and counts at least 4,300 infected routers, a total it says is still rising. The distinction matters. AryStinger exists for the stage of an attack that comes before the break-in. Infected

Security researchers at Paradigm Shift have published a working exploit, dubbed usbliter8, that achieves arbitrary code execution inside the SecureROM of Apple's A12 and A13 chips. That code is burned into the silicon at manufacture. No software update can reach it. Affected devices will carry this flaw for as long as they stay in use. This is not a remote attack. It requires

Microsoft researchers have detailed an exploit chain, named AutoJack, that turns an AI browsing agent into a delivery vehicle for remote code execution. Steer the agent to load an attacker's web page, and that page's JavaScript can reach a privileged local service on the same machine and spawn a process on the host. No credentials, no sign-in screen, and no further user interaction once

Two security teams have shown, in separate research published this week, that OpenClaw, the popular self-hosted AI agent, can be driven to run attacker-controlled code or hand over sensitive data through ordinary-looking inputs. Imperva buried instructions inside shared contacts, vCards, and location pins that the agent executed without the victim ever seeing them. Varonis built a test agent on