This is the first time I've ever seen this.

The German organization HateAid has filed criminal complaints against several managers responsible for the distribution of Meta’s latest smart glasses models in Germany. According to the group, the design of the Meta AI Glasses encourages misuse and potentially violates privacy laws. Under German law, even the sale of disguised spy cameras is prohibited, prompting HateAid to call for intervention by law enforcement to prevent unauthorized recordings. Legal disputes over the glasses have already begun. Hamburg's data protection commissioner, Thomas Fuchs, argues that the warning LEDs on the devices are insufficient to alert users when they are being recorded. He plans to publish a review report on this issue. Meanwhile, the Federal Network Agency appears to hold a different legal interpretation. While the agency can block the import of spy gadgets, such as teddy bears with built-in cameras, it currently does not prohibit any products related to smart glasses. At the heart of the debate lies the question of whether the primary purpose of these glasses is to assist individuals or to facilitate harassment. Meta promotes its glasses as tools that can recommend nearby restaurants or answer questions about the environment, which could benefit visually impaired users who might need assistance reading signs or menus. However, social media discussions focus heavily on another potential use of the glasses: men secretly filming women in public spaces such as airports, supermarkets, fitness studios, or even beaches. These videos often attract derogatory or sexist comments in the comment sections. Meta has responded to these allegations by promising to delete harassment videos from its platforms and implementing a software update to prevent the warning LED from being obscured or manipulated. Josephine Ballon, managing director of HateAid, warns that the increasing popularity of these new types of glasses could create a surveillance pressure. “Whether in the subway, at work, or in the sauna, people are nowhere safe from smart glasses. One must always expect to be filmed and then exposed online,” she explains regarding her legal action. According to the complaint submitted to the SPIEGEL, the design of the latest model is hardly distinguishable from regular Ray-Ban glasses. HateAid's legal action targets not only multiple executives at Meta but also the leadership of the parent company of Ray-Ban and several German retail chains offering the glasses. Whether an investigation will be initiated depends on the Central Office for Combating Internet Crime under the Public Prosecutor's Office in Frankfurt am Main. Short-term inquiries made by the SPIEGEL to the affected companies have yet to receive responses. Women are repeatedly harassed through videos taken with smart glasses. On Instagram, the idea of singing and using copyright law to defend oneself has gone viral. Answers to the most important questions can be found here. Smart glasses represent both a new and old problem. They are nearly indistinguishable from ordinary glasses unless one looks very carefully. Some users not only record their daily lives but also post secret recordings of others online. To combat this abuse, the non-profit organization HateAid filed a criminal complaint on August 12, 2026, seeking to ban the sale of these glasses in Germany. Meta alone sold around seven million smart glasses in 2025, achieving success that eluded Google Glass years ago. A lawyer states that individuals can legally protect themselves against the publication of secretly recorded videos in public spaces. The technical barriers to defense measures are described in an article by the developer of the Smart-Glass Scanner App Nearby Glasses. Experts were consulted about the issue of secret filming with smart glasses. Here, you learn about your rights concerning your image and how to assert them against unwanted filming. The managing director of HateAid explains the background of her criminal complaint. Additionally, the article shows which means of digital self-defense exist against smart glasses. HateAid has filed a criminal complaint against the management of Meta, the eyewear brands Ray-Ban and Oakley, and optical retailers Fielmann, Apollo-Optik, Mister Spex, and MediaMarkt. The sale of smart glasses equipped with integrated cameras violates the Telecommunications and Digital Services Data Protection Act. This law prohibits devices designed to appear as everyday objects for covert filming or eavesdropping. If convicted, fines or prison terms up to two years and confiscation of profits may apply. Particularly targeted is the “Ray-Ban Meta Wayfarer (Gen 2),” which is barely distinguishable from an ordinary sunglasses. Meta uses a blinking LED during recordings, which rarely draws attention and can be manipulated. HateAid demands color coding of cameras, tamper-proof precautions, and clear limits for recordings similar to dash or body cameras. Apple is considering licensing content for Siri AI from publishers. Apple is negotiating long-term licenses with major publishers for its updated voice assistant Siri AI. Unlike previous industry agreements, Apple reportedly does not rely on guaranteed payments but instead on usage-based compensation. The budget is expected to be in the nine-figure range. How publisher content will be presented within the AI system remains unclear. Siri AI will launch with iOS 27 in September. Users in Europe must wait longer due to regulatory reasons. Mistral is expanding its business to become an infrastructure and platform provider. The French Mistral AI is broadening its operations. For the expansion of its own computing capacity in Europe, the company bundles demand from large European firms and institutions that commit to multi-year capacity guarantees. Corporate customers can also decide whether their AI queries should be processed in Europe or the United States. According to VentureBeat, Mistral plans a version where processing runs entirely on its own infrastructure without relying on major cloud providers. The most surprising move: Mistral will offer open-weight models developed by other entities through its platform. It starts with GLM-5.2 from the Chinese AI laboratory Z.ai. Thus, the company earns revenue from foreign models even if customers do not use Mistral models. Strategically, Mistral shifts from a model developer to an infrastructure provider, where computational capacity may become more crucial than individual models. Taiwan reports a cyberattack supported by artificial intelligence on government agencies. Taiwan confirms a cyberattack supported by artificial intelligence originating from abroad targeting state agencies. Countermeasures have been underway since July 20. Which departments were affected and whether data was leaked, the Ministry of Digital Affairs did not disclose. The Israeli cybersecurity firm Dream reports that attackers used freely available AI agents like OpenClaw to build a largely autonomous tool that behaved like a coordinated hacker team. As early as mid-July, up to eight AI agents were simultaneously deployed. Within the networks of 21 state institutions, they searched for vulnerabilities and changed tactics upon encountering resistance. At least 85 government accounts were hacked, and more than 2500 employee records were extracted. Also, Taiwan's nuclear oversight authority and at least seven energy companies became targets of the attack. Simplified Chinese characters in internal communications suggest a perpetrator linked to China according to Dream. Researchers reconstruct user prompts from chatbot answers. From the response of a chatbot, one can deduce what someone asked. A team from the Institute of Technology IIT Bombay and Adobe Research has developed a model that reads a completed AI response and derives the prompt that triggered it.