AI apps with hidden sexual and violent deepfake features made it into the App Store

9to5mac.com Nov 2, 2026

AI apps with hidden sexual and violent deepfake features made it into the App Store

Curated by Shubham · TechCapsules · Nov 2, 2026

AI-summarised brief · reviewed before publication

An investigation by Wired uncovered more than 400 AI‑powered apps, apparently operated by a single Chinese entity, that slipped into both Apple’s App Store and Google Play despite platform policies banning harmful deepfakes. The apps present themselves as benign services—such as language tutoring, pet‑photo editing, or AI workout coaching—but a server‑side check triggered by certain ad clicks activates hidden modules that generate sexual or violent deepfake images and videos from user‑uploaded photos. These modules include scenarios described as “bedroom rape” and “threatening knife.” The network has been promoted through thousands of video ads on Facebook and Instagram, using bait‑and‑switch tactics to pass review. Apple removed 19 apps after the findings; Meta reported disabling over 10,000 ad accounts linked to the domains.

💡 Why It Matters

· The scheme shows how dynamic content swaps can evade static app‑review processes, exposing users to illicit deepfakes even after an app is cleared.

AI Apps Capsules

Read full article on 9to5mac.com

Government Linux Servers Targeted With SysGoRAT and Suspected AI-Generated Scripts

Cybersecurity firm Palo Alto Networks’ Unit 42 tracked a cyber‑espionage campaign against government‑linked Linux servers from October 2025 to April 2026, labeling it CL‑UNK‑1211. Researchers said the attackers used large‑language‑model‑generated PHP and shell scripts, as well as a custom Go‑based remote‑access trojan named SysGoRAT, to gain persistent, multi‑platform access. The malware establishes a WebSocket C2 link to 104.194.158.65:443, disables endpoint defenses, wipes logs and plants fake red‑team markers. Operators exfiltrated database dumps, government mailboxes and social‑media‑related emails, employing in‑memory reverse shells and scripts that enumerate mail directories. The playbook blends open‑source tools, bespoke web shells and AI‑crafted code, allowing rapid adaptation while minimizing manual effort. The campaign also attempted to compress targeted emails with a faulty zip_mails.sh script, illustrating the nature of tooling.

💡 Automating exploit development with LLMs lets adversaries scale attacks on Linux infrastructure that was previously less targeted, reshaping defensive priorities.

thehackernews.com · Nov 2

Is Your SOC Ready for AI Agents? A Maturity Check Before You…

Prophet Security’s 2026 State of AI in Security Operations survey of over 250 CISOs and SOC leaders shows 40 % of security teams use AI daily and 56 % are testing it, with only 4 % planning no adoption. The report outlines a two‑part readiness checklist for deploying AI agents in Security Operations Centers: identifying alert‑handling problems the agents can solve, and ensuring the SOC has the necessary data access, context, written investigation standards, and oversight rules. Technical readiness spans four areas—access and context, baseline metrics, operational rules, and a plan for reclaimed analyst time. The SOC Hierarchy of Needs model emphasizes that most teams struggle at the alert‑management level, where AI agents can immediately improve investigation speed and free analysts for detection, hunting, and posture work. The survey highlights common gaps such as limited tool access, data‑privacy concerns, undocumented investigation standards, and disabled detections, urging a pilot that maps required queries and establishes metrics like investigation share and mean time to investigate.

💡 Without addressing these foundational gaps, AI agents cannot deliver their promised efficiency gains, leaving under‑investigated alerts and wasted analyst capacity.

techcrunch.com · Nov 2

TechCrunch Mobility: A new, bigger AV boom is here

Trucks VC reports a surge in autonomous‑vehicle startups, with 81 pitches in 2026—two‑thirds more than the previous year—indicating a second, larger AV wave. The firm attributes the growth to AI foundation models like Nvidia’s Alpamayo and Qwen’s vision‑language‑action systems, which lower development costs and accelerate deployment. Startups are targeting diverse sectors, from construction to robotaxis, and many position themselves as regional competitors to Waymo. The trend suggests a rapidly expanding, more accessible AV ecosystem.

💡 The influx of AI‑driven AV startups signals a shift toward decentralized, niche autonomy solutions that could reshape regional mobility markets and intensify competition beyond the current industry leaders.

techcrunch.com · Nov 2

AI can make tech CEOs’ lives easier. What about everyone else?

Tech CEOs Marissa Mayer, Sam Altman, and Mark Zuckerberg have sparked backlash by praising AI’s role in everyday life, from school projects to parenting and romance. Their remarks, highlighted by TechCrunch’s Amanda Silberling, raise concerns that industry leaders may be out‑of‑touch with ordinary users. The conversation, featured on the Equity podcast, questions whether AI’s push to replace human interaction is a marketing tactic or a genuine product direction, especially as Anthropic prepares to go public amid cautious investor sentiment.

💡 The CEOs’ statements reveal a disconnect that could erode public trust in AI, prompting scrutiny of how the industry balances innovation with everyday relevance.

techcrunch.com · Nov 2

OpenAI might rebrand its data centers as ‘AI compute campuses’

OpenAI is planning a public‑relations campaign after the November 5 midterm elections, according to Axios. Chief global affairs officer Chris Lehane said the vote will serve as “a movie trailer for what you’re going to see in 2028,” anticipating AI as a central issue in the 2028 presidential race. The campaign will promote several policy ideas: imposing AI “speed limits,” framing AI as an empowerment tool, urging the United States to lead the creation of global AI standards, and rebranding the company’s data‑center facilities as “AI compute campuses.” Lehane praised the ongoing debate over data‑center siting, noting recent moves by Amazon and Microsoft to drop nondisclosure agreements with local governments, and urged communities to demand tangible benefits from such projects.

💡 By reframing data centers as “AI compute campuses,” OpenAI aims to ease local resistance while steering the policy conversation ahead of AI’s rise as a political flashpoint. This strategy could position the firm to influence emerging standards that match its own technology agenda.

appleinsider.com · Nov 2

Sunday Reboot: Sam Altman needs to read the RAM room

Sam Altman, CEO of OpenAI, sparked controversy in a recent interview with the YouTube channel Cheeky Pint by claiming that personal RAM in devices like Macs is wasteful and should be reallocated to cloud AI workloads. He argued that only about 5 % of a consumer’s DRAM is actively used, and that the global shortage of memory—exacerbated by AI’s soaring demand—requires “maximum utilization in the cloud.” Altman’s remarks framed everyday users as impediments to AI progress, suggesting that buying less memory would free resources for large‑scale models. The commentary arrived amid ongoing supply constraints, as memory manufacturers shift production toward high‑bandwidth chips for data centers, leaving limited DRAM for consumer markets.

💡 Altman’s stance spotlights a clash between AI infrastructure needs and consumer rights to upgrade personal hardware, raising questions about who gets priority in a tightening memory market.