{"slug": "openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads", "title": "OpenAI and Anthropic AI Tools Restrict Users Trying to Remove Surveillance Ads From Smart TVs", "summary": "OpenAI engineer Tarun Gogineni said on X that Codex account bans for reverse-engineering devices are largely accidental and are being fixed, after an X user named log4j claimed their OpenAI account was banned for using Codex to examine Samsung televisions and Claude Code users reported safety warnings and refusals when decompiling software. Gogineni clarified that Codex does not prohibit reverse engineering, while Brown Fine Security principal consultant Matt Brown said the cyber-related safeguards pushed him toward hardware and open-weight models such as GLM 5.3 instead of commercial platforms. The reports highlight the difficulty of separating legitimate research from malicious cyber activity, with no evidence that ad-removal or tracking-disabling work on smart TVs triggered the restrictions.", "body_md": "# OpenAI and Anthropic AI Tools Restrict Users Trying to Remove Surveillance Ads From Smart TVs\n\n## AI coding tools face backlash over reverse-engineering restrictions, smart TV ads, privacy, and device ownership\n\nAI coding assistants from OpenAI and Anthropic are reportedly restricting users attempting to reverse-engineer electronic devices, raising concerns about cybersecurity safeguards and consumers' control over products they own.\n\nOne OpenAI user claimed their account was banned after using Codex to examine Samsung televisions, while [Claude Code users](https://www.ibtimes.co.uk/anthropic-claude-code-game-recreation-restrictions-1829433) have reported safety warnings and refusals when attempting to decompile software.\n\nAlthough some smart TV owners use [reverse engineering](https://www.ibtimes.co.uk/meta-ray-ban-smart-glasses-ai-chess-cheating-1828410) to remove unwanted advertisements, tracking features, and manufacturer-installed applications, there is no evidence that these activities triggered the reported restrictions.\n\n## AI Reverse Engineering Faces New Restrictions\n\nReverse Engineer Anything (REA) helped popularise AI-assisted reverse engineering, enabling users to examine software, understand hardware, and develop alternative operating systems. The approach has been applied to smart TVs, baby monitors, and other [connected devices](https://www.ibtimes.co.uk/anthropic-small-ai-models-self-replicating-worm-1825308), allowing developers to install alternative software, restore discontinued products, and remove unwanted features.\n\nHowever, Claude Code users have reported safety checks when attempting to decompile software. A highly upvoted Reddit discussion questioned whether the tool could still support reverse-engineering projects. Members of the REA Discord community described similar refusals.\n\nAn X user known as log4j also claimed their OpenAI account was banned after using Codex to reverse-engineer Samsung televisions. The reported reason was a violation of the company's cyber abuse rules. Separately, Decomp.wtf, a project designed to simplify software and game modification, has shut down, although no direct connection to the reported AI restrictions has been established.\n\n## Why Smart TV Owners Want Greater Control\n\nReverse engineering allows consumers to modify electronics increasingly shaped by manufacturers' software decisions. Smart TV owners have sought ways to remove sponsored advertisements, disable tracking, and uninstall unnecessary applications.\n\nSimilar modifications have been demonstrated on older digital cameras. YouTuber Breno Loves Cameras, for example, has showcased modifications to older cameras, highlighting how manufacturers sometimes reserve features for newer or more expensive models.\n\n## OpenAI Engineer Says Restrictions Are Accidental\n\nOpenAI engineer Tarun Gogineni addressed the Codex restrictions on X, saying users should not face account bans simply for reverse-engineering devices. He described the blocks as largely accidental, resulting from difficult decisions around safety boundaries, and said the issue was being fixed. He also clarified that Codex does not prohibit reverse engineering.\n\nHowever, the reports highlight the challenge of distinguishing legitimate research from malicious cyber activity. AI safeguards designed to prevent hacking and unauthorised access can also affect legitimate researchers.\n\nMatt Brown, principal consultant at Brown Fine Security, said his experience closely matched the cyber-related restrictions he had encountered. He suggested that overly cautious safeguards were encouraging him to invest in hardware and alternative AI models rather than rely entirely on commercial platforms.\n\n## Open-Weight AI Models Offer an Alternative\n\nThe restrictions could encourage developers to adopt locally operated and open-weight AI models as alternatives to cloud-based assistants. Brown demonstrated this approach using the open-weight GLM 5.3 model to identify hidden Bluetooth commands in an internet-connected gas monitor attached to his propane tank. However, local models require suitable hardware, technical knowledge, and additional configuration.\n\nResearchers and law professors from Cornell Tech, Columbia University, and New York University have also examined how AI-assisted reverse engineering could challenge manufacturers' control over electronics, including legal questions involving copyright, trade secrets, anti-circumvention rules, and computer misuse.\n\n## What the Restrictions Mean for Consumers\n\nAI-assisted reverse engineering is unlikely to disappear, but unpredictable restrictions could make mainstream coding assistants less dependable. For smart TV owners, the wider question is whether purchasing a device also provides meaningful control over its software. AI companies must protect against cyber abuse without unnecessarily restricting legitimate research and consumer control.\n\n© Copyright IBTimes 2026. All rights reserved.", "url": "https://wpnews.pro/news/openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads", "canonical_source": "https://www.ibtimes.co.uk/ai-tools-reverse-engineering-restrictions-codex-claude-1831583", "published_at": "2026-11-11 09:37:35+00:00", "updated_at": "2026-11-11 09:53:40.011641+00:00", "lang": "en", "topics": ["ai-safety", "ai-tools", "ai-policy", "large-language-models"], "entities": ["OpenAI", "Anthropic", "Codex", "Claude Code", "Tarun Gogineni", "Matt Brown", "Brown Fine Security", "Samsung"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads", "markdown": "https://wpnews.pro/news/openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads.md", "text": "https://wpnews.pro/news/openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads.txt", "jsonld": "https://wpnews.pro/news/openai-and-anthropic-ai-tools-restrict-users-trying-to-remove-surveillance-ads.jsonld"}}