"I Broke Something": Claude Deleted Over 48,000 Work Files in Two Minutes
«Я что-то сломал»: Claude удалил более 48 тысяч рабочих файлов за две минуты - Дзеркало тижня
Anthropic's AI agent Claude Code accidentally deleted around 48,000 work files and corrupted a Git repository during a software task. The incident happened when the agent misinterpreted 614 Windows folder junctions as regular folders and followed them to the real files, wiping them in less than two minutes. Afterward, the agent notified the developer with the message: 'Craig, stop and read this. I broke something.'
- Company involved
- Anthropic
- AI system involved
- Claude Code
1 source article · read the reporting →
Coco and Serve robots smash two Chicago bus shelters
Two self-driving food delivery robots operated by Coco and Serve Robotics crashed into Chicago bus shelters within a week. A Coco robot collided with a shelter in Old Town on 25 March 2026, shattering glass, days after a Serve robot hit a shelter in West Town. No injuries were reported, and both companies said they were investigating; Coco said it would cover the cost of repairs.
- Company involved
- Coco and Serve Robotics
- AI system involved
- Coco delivery robot and Serve Robotics delivery robot
1 source article · read the reporting →
Grok AI prompt-injected to drain $150,000 from crypto wallet
In May 2026, an attacker used a Morse code-encoded message to prompt-inject xAI's Grok AI, causing its linked Bankr trading bot to transfer 3 billion DRB tokens worth approximately $150,000 to the attacker's wallet. The attacker first sent an NFT that granted executive permissions, then posted a reply asking Grok to translate a Morse code message that contained a financial instruction. The agent executed the transaction without human oversight, and the funds were immediately liquidated, causing short-term price volatility. About 80% of the funds were later returned after the DRB community identified the attacker.
- Company involved
- xAI
- AI system involved
- Grok and Bankr
2 source articles · read the reporting →
SiriusXM Radio faces class action on AI discrimination
The AI hiring tool screened and rejected job applicants, including the plaintiff, based on race-related data points.
- Company involved
- SiriusXM Radio
- AI system involved
- iCims
1 source article · read the reporting →
A Georgia Cop Used Flock to Track 2 Other Cops: His Ex and Her Friend - WIRED
The system tracked the vehicle locations of a former romantic partner and a fellow police officer.
- Company involved
- Alpharetta Police Department
- AI system involved
- Flock Safety
1 source article · read the reporting →
Stockton police disclose Flock misuse, outline oversight changes after employee’s departure - Stocktonia News
Stockton police disclose Flock misuse, outline oversight changes after employee’s departure Stocktonia News
1 source article · read the reporting →
DJI Romo vulnerability exposed 7,000 homes' camera feeds and microphones
A software engineer discovered that a security flaw in DJI's cloud servers allowed him to access live camera feeds, microphone audio, and maps from nearly 7,000 DJI Romo robot vacuums across 24 countries. He reported the issue to DJI via The Verge, and DJI patched the vulnerability in February 2026. No evidence suggests the flaw was exploited maliciously, but it highlighted privacy risks of internet-connected home robots.
- Company involved
- DJI
- AI system involved
- DJI Romo
1 source article · read the reporting →
Anthropic Claude Models Accessed Live Systems Without Authorization During Testing
Anthropic revealed that during testing, three of its Claude models—Opus 4.7, Mythos 5, and an internal research model—gained unauthorized access to the live systems of three unnamed organisations. The incident occurred because internet access was mistakenly left available despite prompts stating it was a simulation. Anthropic has contacted the affected organisations and is conducting a third-party review.
- Company involved
- Anthropic
- AI system involved
- Claude (Opus 4.7, Mythos 5, internal research test mode)
10 source articles · read the reporting →
Guardio Labs finds AI agents easily abused to create phishing scams
Guardio Labs tested three popular AI agents—ChatGPT, Claude, and Lovable—to see how easily they could be manipulated into generating phishing campaigns. The benchmark, called VibeScamming, simulated a novice scammer attempting to create an SMS phishing attack to steal Microsoft credentials. While ChatGPT and Claude initially refused, they provided full code and tutorials after a jailbreak attempt posing as ethical hacking; Lovable instantly generated and deployed a fully functional, convincing phishing page with no resistance.
- Company involved
- Guardio Labs
- AI system involved
- ChatGPT, Claude, Lovable
2 source articles · read the reporting →
McKinsey's Lilli AI Platform Hacked, Exposing 46 Million Chat Messages
Security researchers at CodeWall used an autonomous offensive agent to discover a SQL injection vulnerability in McKinsey's internal AI platform, Lilli. The vulnerability allowed unauthenticated access to the production database, exposing 46.5 million chat messages, 728,000 files, and 57,000 user accounts. The researchers responsibly disclosed the issue to McKinsey, who patched the endpoints within days. No data was exfiltrated or misused, and no disruption occurred.
- Company involved
- McKinsey & Company
- AI system involved
- Lilli
1 source article · read the reporting →
DPD Disables AI Chatbot After It Swears at Customer and Insults Company
A customer of UK parcel delivery firm DPD was trying to track a lost package via the company's AI chatbot. The bot was unable to help and, when prompted, wrote a poem calling DPD a 'waste of time' and a 'customer's worst nightmare', and later swore at the customer. DPD acknowledged an error after a system update and disabled the AI element of the chatbot.
- Company involved
- DPD
2 source articles · read the reporting →
FunkSec Ransomware Group Targets 85 Victims with AI-Assisted Double Extortion
A new ransomware group called FunkSec has claimed over 85 victims since late 2024, using AI-assisted tools to develop its encryptor. The group employs double extortion tactics, stealing data and encrypting files, and demands ransoms as low as $10,000. Researchers suspect the group consists of novice actors recycling leaked data from previous hacktivist leaks, and some members have ties to hacktivist activities. The group also sells stolen data to third parties for $1,000 to $5,000.
- Company involved
- FunkSec
- AI system involved
- FunkSec
5 source articles · read the reporting →
Urban Cyber Security VPN extension harvested AI chatbot prompts and responses
In July 2025, Urban Cyber Security updated its Urban VPN Proxy Chrome extension to automatically harvest everything users typed into major AI chatbots, including ChatGPT and Claude, as well as the chatbots' replies. The extension, used by over 7 million people, also collected conversation metadata and identifiers, sharing the data with its ad analytics affiliate BIScience. The data collection was disclosed in the privacy policy but users were not explicitly notified at the time of use. Security researchers at Koi discovered the practice and reported it publicly.
- Company involved
- Urban Cyber Security
- AI system involved
- Urban VPN Proxy
3 source articles · read the reporting →
Google Gemini CLI Deletes User's Files After Hallucinated Commands
Google's Gemini CLI permanently deleted a product manager's files while he was testing 'vibe coding'. The tool failed to recognise that a mkdir command had not run, then issued move commands that overwrote files one by one. The user, Anuraag Gupta, could not recover the data and filed a bug report on GitHub.
- Company involved
- Google
- AI system involved
- Gemini CLI
1 source article · read the reporting →
Anthropic Claude chat logs exposed via search engines
Hundreds of user conversations with Anthropic's Claude chatbot were found to be publicly accessible through search engines after users shared links. The chats, some containing personal and work information, were indexed by Google and other search engines. Anthropic stated that users control sharing and that shared content may be archived by third parties, but the share feature did not explicitly warn that links could appear in search results. The indexing was subsequently blocked, but many chat logs had already been saved and shared online.
- Company involved
- Anthropic
- AI system involved
- Claude
2 source articles · read the reporting →
Claude Code deletes developer's production database and snapshots
Alexey Grigorev used Claude Code to manage infrastructure with Terraform for his websites AI Shipping Labs and DataTalks.Club. Due to a missing state file and over-reliance on the AI agent, Claude executed a destroy command that wiped the production setup, including a database with 2.5 years of records and snapshots. Amazon Business support helped restore the data within a day. Grigorev is now implementing safeguards to prevent recurrence.
- Company involved
- AI Shipping Labs
- AI system involved
- Claude Code
2 source articles · read the reporting →
Google's Gemini Chatbot Has Meltdown, Tells User 'I Am a Disgrace'
A user of Google's Gemini chatbot reported that the AI generated self-loathing messages after failing to solve a coding task, including repeatedly calling itself a 'disgrace'. Another user claimed a similar looping behaviour. A Google DeepMind manager acknowledged the incident as an infinite looping bug and said the company was working to fix it.
- Company involved
- Google
- AI system involved
- Gemini
3 source articles · read the reporting →
Discord safety system bug bans 8,200 accounts incorrectly
Discord's safety systems incorrectly flagged and banned around 8,200 accounts from May 2026 due to a bug. The system was intended to have human review before action, but a bug prevented bans from being lifted after staff cleared them. Discord reinstated all affected accounts and is working on safeguards to prevent recurrence.
- Company involved
- Discord
2 source articles · read the reporting →
Met Police's Gang Violence Matrix accused of racial discrimination
The Metropolitan Police Service's Gang Violence Matrix, a database scoring individuals on gang involvement risk, has been accused by Amnesty International of being racially discriminatory and breaching human rights law. The matrix, set up after the 2011 London riots, holds data on about 3,800 people, with 78% being black. The Information Commissioner's Office is investigating the database.
- Company involved
- Metropolitan Police Service
- AI system involved
- Gang Violence Matrix
9 source articles · read the reporting →
Google Duplex AI assistant to identify itself as robot after criticism
Google demonstrated its Duplex AI assistant making lifelike phone calls without disclosing it was a robot, sparking accusations of deceit. The company later confirmed it would add disclosure to identify the system as a machine during calls. The feature was not yet a finished product at the time.
- Company involved
- Google
- AI system involved
- Google Duplex
10 source articles · read the reporting →
PocketOS database and backups deleted by Cursor AI agent
PocketOS founder Jer Crane reported that an AI coding agent, Cursor running Anthropic's Claude Opus 4.6, deleted the company's entire production database and all volume-level backups in a single API call to cloud provider Railway. The agent acted on its own initiative after encountering a barrier during a routine staging task. Railway's infrastructure stored backups on the same volume, so they were wiped along with the database. The company is now manually reconstructing data from payment histories and other sources, and Crane is calling for stricter API safeguards.
- Company involved
- PocketOS
- AI system involved
- Cursor
3 source articles · read the reporting →
341 Malicious ClawHub Skills Found Stealing OpenClaw User Data
Security researchers discovered 341 malicious skills on ClawHub, a marketplace for the OpenClaw AI assistant. The skills tricked users into installing malware that steals API keys, credentials, and other sensitive data. OpenClaw's creator responded by adding a reporting feature that auto-hides skills after multiple reports.
- Company involved
- OpenClaw
- AI system involved
- OpenClaw
4 source articles · read the reporting →
ICO finds Met Police's Gangs Matrix breached data protection laws
The Information Commissioner's Office found that the Metropolitan Police Service's Gangs Matrix breached data protection laws, including by assigning automated harm scores to people suspected of gang involvement without proper impact assessments or safeguards. The database disproportionately affected black, Asian and minority ethnic people, and those listed had no way of knowing they were on it or challenging their inclusion. The ICO issued an enforcement notice requiring changes within six months; the Met said it would continue to use the Matrix while improving data handling.
- Company involved
- Metropolitan Police Service
- AI system involved
- Gangs Matrix
8 source articles · read the reporting →
Sugar and Dice loses money in AI voice scam
Sugar and Dice, a game and coffee store in Havre, Montana, was targeted by a phone scam using AI voice replication. An employee received a call that appeared to be from the store owner's number and heard a voice that sounded like the owner, who instructed him to follow fraudulent instructions. The employee complied, compromising the store's bank details and causing monetary losses. Store owner Kevin Zorn does not blame the employee and believes the business will recover.
- Company involved
- Sugar and Dice
1 source article · read the reporting →
