OpenAI Disables Coordinated ChatGPT Network Used for Financial Scams and Identity Forgery
OpenAI has deactivated a coordinated network of ChatGPT accounts used to support financial scams, romance fraud, and identity forgery operations.
Criminals employed the AI system to create fake personas, translate conversations in real time, and generate personalized messages directed at potential victims. The investigation began after OpenAI received reports of suspicious activity on WhatsApp.
Among the schemes were fraudulent investments in cryptocurrencies and gold, romance-based confidence tricks, fake bonus offers from betting platforms, and demands for payments issued under the guise of police authorities.
ChatGPT was also used to produce forged documents such as stock purchase confirmations, legal notices, passports, and fabricated interfaces of financial services. These materials helped make the fraudulent narratives more convincing to targets.
Attacks generally started with contact via social media or messaging applications. After establishing emotional bonds or creating a sense of urgency, operators requested deposits, activation fees, nonexistent fines, or payments to release supposed rewards.
The investigation also identified possible indicators of human trafficking and forced labor. Accounts generated job advertisements for “attendants” in Poipet and discussed debts, disciplinary penalties, worker retention, expired visas, and escape attempts.
All involved accounts have been blocked. OpenAI has shared operational indicators with law enforcement agencies and technology companies.
Related articles
Why Isolated Bank Scoring Fails to Prevent Counterparty Default: OSINT Case Study of a Scam Deal
A Russian social entrepreneur lost nearly one million rubles after relying on a major bank's automated compliance system that assigned a green reliability score to a fraudulent supplier. The supplier, registered as a transport company, received 500,000 rubles in targeted government grant funds for IT equipment but delivered only 15,000 rubles worth of goods before disappearing. Retrospective analysis using public Russian registries revealed multiple red flags missed by the isolated scoring algorithm, including mismatched OKVED codes, missing financial statements for three years, and clear ownership links to a bankrupt IT firm. The beneficial owner later faced tax blacklist entry and three-year disqualification after the transport entity was liquidated as inactive. The case demonstrates that single-entity automated checks create a false sense of security by ignoring beneficiary networks and operational logic. Manual verification through official sources such as EGRUL, GIR BO, and arbitration court records proved far more effective at exposing the scheme.
Avito and BI.ZONE Launch Cyberdetective CTF for Non-Technical Employees Focused on Phishing and Social Engineering
Avito's Security Awareness team, in partnership with BI.ZONE, created the Cyberdetective jeopardy-style CTF competition specifically for non-technical staff who do not write code. The three-day event targeted 104 registered participants, with 44 completing tasks and 16 achieving maximum scores across scenarios involving phishing, OAuth consent abuse, file metadata analysis, and prompt injection. Tasks awarded flags not only for investigation but also for refusing risky actions such as granting excessive application permissions or clicking suspicious links. The design incorporated MITRE ATT&CK techniques including T1566.002 Spearphishing Link, T1528 Steal Application Access Token, and T1204.004 Malicious Copy and Paste. Organizers emphasized indirect hints to help participants recognize when an object required action, avoiding the common barrier where employees fail to understand that a strange string or domain needs examination. Feedback highlighted the value of hands-on decision-making over passive lectures, though some technical employees completed the set rapidly.
Paper Mail Combined with Phishing: Russians Targeted by Fake FNS Tax Portal Copies
Russian scammers have started blending traditional postal mail with digital phishing to steal taxpayer credentials. Victims first receive official-looking paper letters from the Federal Tax Service demanding corrections to tax declarations or supporting expense documents. These letters are followed by emails directing users to a counterfeit version of the FNS personal account portal. The fake site captures login details, passwords, and other authentication data that can be reused for further identity theft and account takeovers. Senator Artem Sheikin highlighted the scheme, noting that the paper letter builds initial trust before the electronic follow-up appears legitimate. Users are advised to access tax information only by manually visiting the official FNS website or through the Gosuslugi platform without clicking any links. The attack relies primarily on psychological manipulation rather than technical sophistication.
The Human Factor in Phishing: Why Employees Wait Nine Days to Report Credential Theft
A detailed case analysis reveals how an employee clicked a phishing link, entered credentials, and realized the mistake within a minute, yet security teams only discovered the incident nine days later through an automated rule rather than a human report. The delay stemmed from immediate shame and motivated reasoning that allowed the victim to convince themselves no further action was needed, including the false belief that changing the password resolved the issue. This nine-day window left attacker sessions active, permitted potential mailbox forwarding rules to exfiltrate data, and risked rotating logs that erased forensic evidence. The article examines how corporate cultures that frame incidents as blameworthy events actively discourage reporting and turn minor phishing successes into prolonged investigations involving third parties. Recommendations include creating a no-blame reporting policy with clear boundaries, implementing one-click in-client reporting buttons, and shifting metrics from click rates to reporting speed and volume. The analysis stresses that even strong technical detection cannot fully replace rapid human signals when cultural barriers remain unaddressed.