{"id":"CMP-202611-uh4cao","kind":"campaign","aliases":[],"related":[],"title":"Russian-speaking attacker runs 'exploit_agent', an AI agent that turns new web app security patches into working exploits with no human in the loop, and steals data from French media, entertainment and education websites (Google GTIG)","description":null,"summary":"Google Threat Intelligence Group's Q4 2026 AI Threat Tracker (16 Nov 2026) describes exploit_agent, a multi-agent framework that works with several frontier and local LLMs. GTIG found it on an exposed staging server of a suspected Russian-speaking operator. It keeps comparing the code of popular web application software before and after each security update, works out the flaw, writes an exploit and runs it automatically, with no human involved. GTIG calls this the first fully autonomous discovery-to-exploitation it has seen. The server held data stolen from victims and scripts for internal reconnaissance, brute-forcing, installing webshells and removing other attackers' webshells. Victim telemetry points to French websites in media, entertainment and education. GTIG has not publicly named which web applications were targeted, the victims, or the dates. The same report says attackers also use the ARTEX, CAIRN, CyberStrikeAI, HexStrike AI and PentAGI AI frameworks (PentAGI with proxies against aerospace). The point for defenders: the time between a web app patch and exploitation is getting much shorter.","severity":"high","cvss":null,"epss":null,"exploited":true,"exploited_at":"2026-11-16T14:36:53+00:00","products":[],"ecosystems":[],"tags":["campaign","early-warning"],"not_affected_if":[],"how_to_check":"- You run internet-facing websites built on popular web application software (CMS, forums, e-commerce, LMS). The specific products are unknown. Higher risk if you are a French media, entertainment or education site.\\n- Look for webshells or unexpected PHP/script files in web roots, and for webshells that disappeared (the operator deletes rival webshells).\\n- Look for brute-force bursts against admin logins, and for exploit-like requests hitting endpoints changed in a recent security update soon after it came out.\\n- Indicators (IPs, domains, hashes): GTIG has not published any in the coverage so far; unknown.","fix":"- Apply security updates for internet-facing web applications and their plugins within hours to days, not weeks. Assume exploits can exist as soon as the patch is public.\\n- Use a WAF or virtual patching for apps you cannot update quickly.\\n- Require MFA and lockout on admin logins.\\n- Monitor web roots for new files and investigate any webshell. If you find one, treat the site and its stored data as compromised and rotate credentials.","status":"open","published_at":null,"first_seen_at":"2026-11-16T14:36:53+00:00","triaged_at":"2026-11-16T14:36:53+00:00","updated_at":"2026-11-16T14:38:05+00:00","category":"infrastructure","confidence":null,"triage":{"cause":{"at":"2026-11-16T17:22:10Z","path":"none","entry":"unpatched-vuln","reason":"exploit_agent watches web application security patches and automatically writes and runs exploits for the flaws they fix, and it steals data from French media, entertainment and education websites, but GTIG has not said which applications or victims were hit or how each intrusion began.","history":[{"at":"2026-11-16T14:37:51Z","path":"unknown","entry":"unpatched-vuln","confidence":0.6}],"confidence":0.5}},"analyst_status":"done","analyst_note":null,"search":"'-202611':40A '16':56B '2026':52B,58B 'aerospac':204B 'agent':7A,10A,61B,65B 'ai':9A,53B,195B,198B 'also':188B 'app':15A,214B 'applic':98B,175B 'artex':191B 'attack':4A,155B,187B 'automat':116B 'autonom':127B 'brute':148B 'brute-forc':147B 'cairn':192B 'call':122B 'campaign':42B 'cmp':39A 'code':94B 'compar':92B 'cyberstrikeai':193B 'data':29A,138B 'date':182B 'defend':208B 'describ':59B 'discoveri':129B 'discovery-to-exploit':128B 'earli':44B 'early-warn':43B 'educ':35A,167B 'entertain':33A,165B 'exploit':6A,20A,60B,112B,131B,217B 'expos':80B 'first':125B 'flaw':109B 'forc':149B 'found':76B 'framework':66B,199B 'french':31A,161B 'frontier':71B 'fulli':126B 'get':219B 'googl':37A,46B 'group':49B 'gtig':38A,75B,121B,168B 'held':137B 'hexstrik':194B 'human':23A,119B 'instal':150B 'intellig':48B 'intern':145B 'involv':120B 'keep':91B 'llms':74B 'local':73B 'loop':26A 'media':32A,164B 'much':220B 'multi':64B 'multi-ag':63B 'name':172B 'new':13A 'nov':57B 'oper':89B 'patch':17A,215B 'pentagi':197B,200B 'point':159B,206B 'popular':96B 'proxi':202B 'public':171B 'q4':51B 'reconnaiss':146B 'remov':153B 'report':185B 'run':5A,114B 'russian':2A,87B 'russian-speak':1A,86B 'say':186B 'script':143B 'secur':16A,104B 'seen':134B 'server':82B,136B 'sever':70B 'shorter':221B 'softwar':99B 'speak':3A,88B 'stage':81B 'steal':28A 'stolen':139B 'suspect':85B 'target':177B 'telemetri':158B 'threat':47B,54B 'time':210B 'tracker':55B 'turn':12A 'uh4cao':41A 'updat':105B 'use':189B 'victim':141B,157B,179B 'warn':45B 'web':14A,97B,174B,213B 'webshel':151B,156B 'websit':36A,162B 'work':19A,68B,106B 'write':110B","nicknames":["exploit_agent"],"mitigations":[],"who_affected":null,"tier":"A","verified":{},"analyst_revisit_at":null,"fix_status":null,"audience":null,"plain_action":null,"pkeys":["exploitagent"],"ptokens":[],"labels":["campaign","earlywarning","high","infrastructure"],"campaign_type":"exploitation","search_a":"'-202611':40A 'agent':7A,10A 'ai':9A 'app':15A 'attack':4A 'cmp':39A 'data':29A 'educ':35A 'entertain':33A 'exploit':6A,20A 'french':31A 'googl':37A 'gtig':38A 'human':23A 'loop':26A 'media':32A 'new':13A 'patch':17A 'run':5A 'russian':2A 'russian-speak':1A 'secur':16A 'speak':3A 'steal':28A 'turn':12A 'uh4cao':41A 'web':14A 'websit':36A 'work':19A","search_b":"'16':56B '2026':52B,58B 'aerospac':204B 'agent':61B,65B 'ai':53B,195B,198B 'also':188B 'app':214B 'applic':98B,175B 'artex':191B 'attack':155B,187B 'automat':116B 'autonom':127B 'brute':148B 'brute-forc':147B 'cairn':192B 'call':122B 'campaign':42B 'code':94B 'compar':92B 'cyberstrikeai':193B 'data':138B 'date':182B 'defend':208B 'describ':59B 'discoveri':129B 'discovery-to-exploit':128B 'earli':44B 'early-warn':43B 'educ':167B 'entertain':165B 'exploit':60B,112B,131B,217B 'expos':80B 'first':125B 'flaw':109B 'forc':149B 'found':76B 'framework':66B,199B 'french':161B 'frontier':71B 'fulli':126B 'get':219B 'googl':46B 'group':49B 'gtig':75B,121B,168B 'held':137B 'hexstrik':194B 'human':119B 'instal':150B 'intellig':48B 'intern':145B 'involv':120B 'keep':91B 'llms':74B 'local':73B 'media':164B 'much':220B 'multi':64B 'multi-ag':63B 'name':172B 'nov':57B 'oper':89B 'patch':215B 'pentagi':197B,200B 'point':159B,206B 'popular':96B 'proxi':202B 'public':171B 'q4':51B 'reconnaiss':146B 'remov':153B 'report':185B 'run':114B 'russian':87B 'russian-speak':86B 'say':186B 'script':143B 'secur':104B 'seen':134B 'server':82B,136B 'sever':70B 'shorter':221B 'softwar':99B 'speak':88B 'stage':81B 'stolen':139B 'suspect':85B 'target':177B 'telemetri':158B 'threat':47B,54B 'time':210B 'tracker':55B 'updat':105B 'use':189B 'victim':141B,157B,179B 'warn':45B 'web':97B,174B,213B 'webshel':151B,156B 'websit':162B 'work':68B,106B 'write':110B","search_c":"","cause_entry":"unpatched-vuln","cause_path":"none","disclosures_checked_at":null,"first_public_at":"2026-11-16T00:00:00+00:00","kev":null,"cvss_vector":null,"cvss_by":null,"poc_count":0,"sources":[{"source":"analyst","external_id":"CMP-202611-uh4cao","url":"https://www.theregister.com/research/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-no-human-help/5309926","severity":null,"published_at":null,"modified_at":null,"first_seen_at":"2026-11-16T14:36:53+00:00","fetched_at":"2026-11-16T14:36:53+00:00","note":null,"cites":["https://www.theregister.com/research/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-no-human-help/5309926"]}],"documents":[{"id":169884,"url":"https://x.com/TheHackersNews/status/2122330337334464721","title":"@TheHackersNews: 🛑 Google warns hackers are using autonomous AI agents to turn security patches i","kind":"post","source":"x","publisher":null,"text":"🛑 Google warns hackers are using autonomous AI agents to turn security patches into working exploits.\n\nAttackers are deploying multi-agent frameworks to automate hacking, weaponize known flaws, and speed up intrusions.\n\nHere's what Google uncovered: https://t.co/4LYuPonBzW\n\nlikes 246 reposts 68","published_at":"2026-11-16T14:05:58+00:00","relation":"describes","confidence":null,"note":null},{"id":169861,"url":"https://x.com/BleepinComputer/status/2122328904836829643","title":"@BleepinComputer: Google warns that fully-automated AI attacks are coming soon\nhttps://t.co/Rst8y3","kind":"post","source":"x","publisher":null,"text":"Google warns that fully-automated AI attacks are coming soon\nhttps://t.co/Rst8y3wKOm\n\nhttps://t.co/Rst8y3wKOm\n\nlikes 109 reposts 38","published_at":"2026-11-16T14:00:16+00:00","relation":"describes","confidence":null,"note":null},{"id":169949,"url":"https://news.ycombinator.com/item?id=50517797","title":"Google: AI agent turned a security patch into a exploit with no human help","kind":"post","source":"hackernews","publisher":null,"text":"Google: AI agent turned a security patch into a exploit with no human help\nhttps://www.theregister.com/research/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-no-human-help/5309926\n\n2 points, 2 comments on Hacker News","published_at":"2026-11-16T14:14:09+00:00","relation":"describes","confidence":null,"note":null},{"id":169867,"url":"https://www.bleepingcomputer.com/news/security/google-warns-that-fully-automated-ai-attacks-are-coming-soon/","title":"Google warns that fully-automated AI attacks are coming soon","kind":"blog","source":"blog","publisher":null,"text":"GTIG AI Threat Tracker: exploit_agent fully automated vuln finding/exploitation against French media, entertainment, education websites; analyzes software updates of popular web apps. UNC6836 (suspected links to Midnight Neptune, North Korea) targets cryptocurrency firms in North America and Europe using modified Claude Code: after initial compromise, attackers modified the Claude Code app's settings so that at launch it triggered download of the SOMBERMEME backdoor. ITmedia adds UNC6836 approached targets with AI-generated fake recruiter profiles.","published_at":"2026-11-16T00:00:00+00:00","relation":"describes","confidence":null,"note":null},{"id":169979,"url":"https://www.theregister.com/research/2026/11/16/google-ai-agent-turned-a-security-patch-into-a-working-exploit-with-no-human-help/5309926","title":"Google: AI agent turned a security patch into a working exploit with no human help","kind":"blog","source":"analyst","publisher":null,"text":"GTIG Q4 2026 AI Threat Tracker: suspected Russian-speaking operator ran exploit_agent, a multi-agent framework integrating frontier and local LLMs, found on an exposed staging server. It continuously diffs popular web application software before/after security updates, finds flaws, writes exploits and runs them with no human. Server held stolen victim data and post-compromise scripts for internal recon, brute-forcing, installing webshells and removing other attackers' webshells. Victim telemetry: French websites in media, entertainment and education. Other frameworks abused: ARTEX, CAIRN, CyberStrikeAI, HexStrike AI, PentAGI (PentAGI with proxies vs aerospace).","published_at":"2026-11-16T00:00:00+00:00","relation":"describes","confidence":null,"note":null},{"id":171370,"url":"https://x.com/GoogleCloudSec/status/2122369113594966244","title":"@GoogleCloudSec: New GTIG AI Threat Tracker:\n\nAs 2027 nears, human latency is leaving the attack ","kind":"post","source":"x","publisher":null,"text":"New GTIG AI Threat Tracker:\n\nAs 2027 nears, human latency is leaving the attack chain—from autonomous patch-to-exploit pipelines to hijacked cloud quotas &amp; backdoored configs.\n\n📄 Full report: https://t.co/lNcURijizF\n\n▶️ 2027 insights: https://t.co/fBbsskkBuG https://t.co/uuiJN76nZd\n\nlikes 11 reposts 5","published_at":"2026-11-16T16:40:03+00:00","relation":"discussion","confidence":null,"note":null}],"disclosures":[],"exploit_sources":[],"notice":"Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai"}