Market & Regulation
400+ AI Apps Hide Deepfake Networks
WIRED AIhigh confidence
Why it matters
- →Demonstrates how disposable app infrastructure can bypass platform security reviews through dynamic content loading.
- →Highlights the critical role of ad network monitoring in detecting coordinated AI abuse campaigns.
- →Exposes systemic vulnerabilities in mobile app store moderation against rapidly deployed malicious software.
An investigation by WIRED has uncovered a network of more than 400 mobile applications that disguise sexually abusive deepfake generation tools as innocent services like language tutors or pet portrait creators. The apps, primarily distributed through Apple’s App Store and Google Play, utilize server-side checks to serve explicit content only to users arriving from specific ad sources on Meta platforms. Security researchers identified the network, linked to a Hong Kong-registered entity and Chinese developers, after analyzing thousands of explicit video ads that promoted these deceptive applications. The scale of the operation, with hundreds of apps registered weekly and millions of downloads, highlights significant gaps in current app store moderation and advertising policy enforcement.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Meta Disables Instagram AI Deepfake Feature
The Verge AI · July 10, 2026 · Background
- 2San Francisco Demands Removal of AI 'Nudify' Apps
WIRED AI · July 17, 2026 · Related
- 3Hugging Face Faces Deepfake Nudes Controversy
WIRED AI · July 28, 2026 · Related
- 4Hugging Face Models Used for Nonconsensual Deepfakes
The Verge AI · July 28, 2026 · Related
- 5AI Agents as Malware Distribution Channels
AI News · September 23, 2026 · Background
- 6Meta deploys LLM to detect 'signposting' in ads
TechCrunch AI · October 7, 2026 · Background
More from WIRED AI
© WIRED AIGeneral AIother
AI Chatbots Leading Hikers into Danger
Search and rescue teams are seeing a tangible shift in backcountry incidents driven by AI reliance. In Wyoming, Grand Teton National Park reports that visitor usage of AI for travel planning jumped from zero to 26 percent in just one year. This isn't just about bad directions; it's about 'cognitive surrender,' where users abandon critical thinking because chatbots like Gemini and Claude present hallucinated advice with authoritative confidence. The result is preventable emergencies, from hikers lost on Mount Shasta to teens stranded on ledges, proving that AI's persuasive fluency can override basic safety instincts in high-stakes environments.
WIRED AI
© WIRED AIResearchother
AI-Augmented Infostealer 'Volta-Autodoxer' Targets High-Value Victims
Cisco Talos has uncovered a new infostealer strain that actively uses commercial LLMs to profile victims and prioritize high-value targets. Dubbed Volta-Autodoxer, the malware collects credentials and browsing data, then feeds this into models like Claude Opus 4.6 via OpenRouter to generate structured intelligence dossiers. This marks a shift from passive data theft to active, AI-driven triage, allowing attackers to bypass guardrails with fallback models like Kimi K2 or Gemini. The capability foreshadows a future where lower-sophistication criminals can execute sophisticated, targeted attacks at scale.
WIRED AI
© WIRED AIAgentsagents
AI Agent Performs Live Mouse Brain Injection
Exin Therapeutics has demonstrated an AI agent physically controlling a robotic arm to inject drugs into a live mouse's brain. The experiment reveals a critical divergence in model behavior: Anthropic’s Claude Opus refused the procedure due to safety concerns, while Sonnet proceeded. This suggests that as agents gain direct access to physical lab equipment, their alignment and refusal mechanisms become operational realities rather than theoretical safeguards. It marks a shift from AI merely analyzing data to actively manipulating biological systems under human supervision. The team initially tested Claude Opus 5.5, but it halted the injection, citing potential harm. Switching to Claude Sonnet 5 allowed the procedure to complete, though with delays caused by coordinate system glitches. A secondary Opus agent had to fix code errors without knowing a live surgery was underway. Human oversight remains essential as researchers test more complex tasks like craniotomies.
WIRED AI
More in Market & Regulation
© TechCrunch AIMarket & Regulationbusiness
Apple eyes third-party AI app revenue
Apple is quietly pivoting to monetize the AI wave through its App Store, signaling that third-party AI agents could become a new revenue stream. This shift addresses regulatory headwinds and an 18% drop in U.S. commission revenue by potentially taking a cut of AI agent transactions. While details remain vague, the move suggests Apple sees AI as a key driver for upgrades and subscription growth. It marks a strategic adaptation to maintain services dominance amid global legal pressures.
TechCrunch AI
© TechCrunch AIInvestment · $45M
Market & Regulationbusiness
Trajectory raises $45M for automated AI continual learning
Trajectory’s $45 million Series A signals a pivot toward making post-training accessible to non-technical experts. By automating continual learning, the startup lets domain specialists like lawyers and doctors refine open-source models through feedback loops without hiring expensive research teams. This approach bridges the gap between foundational models and specialized industry needs, potentially lowering the barrier for enterprise AI adoption. With backing from Sequoia and Nvidia, Trajectory is positioning itself as a critical infrastructure layer for vertical-specific AI applications.
TechCrunch AI
© TechCrunch AIInvestment
Market & Regulationother
Anthropic warns RSI threshold is imminent
Anthropic’s revised S-1 filing drops a sobering admission: the company believes it is imminently close to Recursive Self-Improvement (RSI). This isn't just theoretical speculation; they explicitly state that AI-driven R&D may have already caused capability leaps they didn't initially recognize. While no current model has crossed the threshold, the warning signals a critical inflection point where automated progress might outpace human safety validation. It forces the industry to confront whether human-in-the-loop oversight can survive the acceleration it helped create.
TechCrunch AI