News/Privacy
New AI apps are showing deepfake videos of people kissing without consent
Meta and TikTok have been displaying thousands of advertisements for AI applications that generate synthetic videos of people kissing without consent, raising serious ethical concerns about the normalization of deepfake technology. Current situation: Social media platforms have been running numerous advertisements for AI apps that can create artificial videos of any two people kissing, based solely on uploaded photos. Meta has shown over 2,500 ads across Instagram and Facebook, with approximately 1,000 currently active TikTok has displayed around 1,000 ads to millions of European users Many ads feature celebrities like Scarlett Johansson and Emma Watson, while others promote the ability...
read Jan 17, 2025The FTC is escalating concerns over Snap’s AI chatbot to the DOJ
The Federal Trade Commission (FTC) has escalated concerns about Snap's AI chatbot to the Department of Justice (DOJ), citing potential privacy violations related to a 2014 settlement agreement. Key developments: The FTC's investigation into Snap's compliance with a decade-old privacy settlement has led to serious concerns about the company's My AI chatbot feature. The agency has indicated it has found evidence suggesting Snap is either currently violating or on the verge of violating privacy laws This referral to the DOJ represents a significant escalation in regulatory scrutiny of AI-powered social media features The investigation specifically focuses on the My AI...
read Jan 14, 2025AI assistants will gain access to your screen in 2025
AI assistants now integrate directly with various applications, simplifying tasks and offering contextual help through visual recognition and natural language interaction. Key innovations: Nvidia's Project RTX and Razer's Project AVA represent significant advances in AI assistance technology, demonstrating how AI can actively monitor and respond to user activities in real-time. Project RTX features a graphical assistant that observes screen activity and webcam input to provide contextual help and execute commands The system can identify applications being used and guide users through specific features, such as Photoshop's Generative Fill Physical object recognition through webcam enables real-world interactions, like identifying a coffee...
read Jan 9, 2025Apple defends data privacy practices amid Siri lawsuit
Key Development: Apple has responded to privacy concerns raised in a lawsuit about Siri by releasing detailed information about its data protection practices and settling the case regarding third-party grading concerns from 2019. Core Privacy Claims: Apple maintains a strict policy regarding Siri data usage and user privacy protection. The company explicitly states it has never used Siri data to create marketing profiles Siri data has never been made available for advertising purposes No user data has been sold to third parties for any purpose Technical Implementation: Apple prioritizes on-device processing to minimize data exposure and protect user privacy. Personalized...
read Jan 9, 2025Cohere just launched ‘North’ in a big bet to bring AI to privacy-focused enterprises
Microsoft's Copilot and Google Vertex AI face new competition as Cohere launches North, a secure AI workspace platform designed specifically for enterprise deployment in regulated industries. The platform overview: North combines large language models, search capabilities, and automation tools in a secure environment that prioritizes data privacy and control. The platform can be deployed in private cloud environments or on-premises installations North includes Compass, a built-in search system that processes multiple data types including images, presentations, spreadsheets, and documents across languages Internal testing indicates Compass reduces task completion times by over 80% compared to manual searches Performance metrics: Cohere's internal...
read Jan 6, 2025Apple is prioritizing user privacy protection over winning the AI race
Apple's strategic approach to AI development prioritizes user privacy and security while focusing on practical applications, marking a distinct contrast to competitors' race for generative AI dominance. Core strategy and philosophy: Apple maintains a fundamentally different approach to AI development, emphasizing privacy as a human right and prioritizing on-device processing over cloud-based solutions. The company processes AI features like text prediction, Face ID, and image recognition locally on devices Siri handles many tasks without transmitting data to external servers, preserving user confidentiality This privacy-first stance creates technical limitations compared to competitors who leverage vast cloud resources Current AI implementation: While...
read Jan 3, 2025Apple reportedly opts everyone into having their personal photos analyzed by AI
Apple introduced automated photo landmark identification capabilities in iOS and macOS through Enhanced Visual Search, activating the feature by default across its ecosystem. Key Implementation Details: Apple's Enhanced Visual Search leverages AI and machine learning to analyze users' photos for landmarks and points of interest, processing this data through both local and cloud-based systems. The feature was quietly rolled out with iOS 18.1 and macOS 15.1 updates on October 28, 2024 The system employs advanced privacy technologies including homomorphic encryption (which allows data processing while remaining encrypted) and differential privacy techniques Photo analysis occurs locally on devices before encrypted data...
read Jan 3, 2025Apple to pay $95M in Siri privacy lawsuit settlement
Apple has agreed to a $95 million settlement following allegations that Siri recordings violated user privacy between 2014 and 2024. Key details of the settlement; The class action lawsuit, filed in 2019, covers customers who used Siri-enabled Apple devices over a ten-year period. The settlement awaits approval from U.S. District Judge Jeffrey White in Oakland, California Affected customers could receive up to $20 in compensation, though the final amount may be lower after legal fees Legal teams can seek up to $28.5 million in fees and $1.1 million for expenses Core allegations: Users claimed that Apple's voice assistant recorded private...
read Jan 3, 2025Apple’s new photo feature captures more data than you may want– here’s how to turn it off
Apple recently launched Enhanced Visual Search (EVS), a photo recognition feature in iOS 18 that has sparked privacy concerns due to its default-enabled data sharing with Apple's servers. Key functionality: Enhanced Visual Search uses on-device AI to identify landmarks and points of interest in photos, creating mathematical fingerprints that are sent to Apple's servers for matching. The system generates "vector embeddings" for detected points of interest in photos Data is transmitted through a privacy relay system that includes fake queries to mask genuine requests Matches are returned to users in encrypted form Privacy concerns: Developer Jeff Johnson has highlighted that...
read Dec 29, 2024How ‘federated learning’ in AI enhances privacy without sacrificing innovation
Federated learning represents a significant advancement in AI technology that enables machine learning models to learn from distributed data sources while maintaining data privacy and security. Core concept and innovation: Federated learning fundamentally changes how AI systems learn by bringing the model to the data rather than centralizing data in one location, enabling privacy-preserving machine learning at scale. Instead of collecting data in a central repository, the AI model travels to where data resides, whether on smartphones, hospital servers, or smart devices The approach allows AI systems to learn from millions of data points while keeping sensitive information secure at...
read Dec 27, 2024Law enforcement agencies are turning to AI to catch human traffickers
A new study examines how artificial intelligence and deep learning technologies are being deployed to combat human trafficking through advanced detection and surveillance systems. The foundation of AI-powered trafficking detection: Deep learning, a subset of artificial intelligence that uses complex neural networks to analyze large datasets, is emerging as a powerful tool in identifying patterns associated with human trafficking activities. Deep learning algorithms can be trained to recognize specific indicators of trafficking, including suspicious financial transactions, unusual travel patterns, and distinctive language in online advertisements The technology's ability to process and analyze vast amounts of data makes it particularly effective...
read Dec 20, 2024OpenAI hit with €15M fine by Italy for privacy violations
Italy's data protection authority has imposed a significant fine on OpenAI for privacy violations related to ChatGPT, marking a major regulatory action against the AI company in Europe. Key enforcement action: Italy's data protection agency has fined OpenAI 15 million euros ($15.58 million) following an investigation into ChatGPT's handling of personal data. The regulator found that OpenAI processed users' personal data to train ChatGPT without proper legal basis The company failed to meet transparency requirements and information obligations to users OpenAI lacked adequate age verification systems to protect children under 13 from inappropriate AI-generated content Regulatory requirements: The Italian watchdog...
read Dec 18, 2024EU authorities open door to AI using personal data without consent
The European Data Protection Board's latest guidance explores how companies can develop AI models while adhering to data privacy regulations, particularly focusing on the use of personal data in training processes. Key framework developments: The EDPB's new report outlines potential pathways for using personal data in AI training without explicit consent, marking a significant shift in regulatory thinking. The guidance suggests that personal data could be used for AI training if the final application does not reveal private information about individuals This interpretation acknowledges the technical distinction between training data and the information ultimately delivered to end users The framework...
read Dec 16, 2024UnitedHealth AI glitch exposes claims-judging tool to public
The healthcare industry continues to grapple with AI implementation challenges as UnitedHealth Group faces scrutiny over an accidentally exposed claims-processing chatbot. The security breach: A chatbot used by UnitedHealth's Optum Rx pharmacy benefit manager to process insurance claims and disputes was inadvertently made public and accessible to anyone with its IP address. The exposed system, called "SOP Chatbot," was designed to handle standard operating procedure queries for employees Employee interaction logs revealed questions about policy renewal dates and claim determinations Cybersecurity researcher Mossab Hussein, co-founder of spiderSilk, discovered and reported the privacy breach Company response: UnitedHealth quickly locked down access...
read Dec 14, 2024Lawsuit claims Photobucket sold user biometric data illegally
Photobucket faces a class action lawsuit over allegations of selling users' biometric data to AI companies without proper consent, potentially affecting up to 100 million users and billions of photos stored on the platform since 2003. Key allegations and scope: The lawsuit targets Photobucket's recent privacy policy update that revealed plans to sell users' photos, including biometric data like face and iris scans, to AI training companies. Two distinct classes are represented: users who uploaded photos between 2003 and May 2024, and non-users whose images appear in uploaded photos The company claimed access to approximately 6.5 billion public images eligible...
read Dec 9, 2024Gmail uses AI to summarize your emails, but should it?
Core functionality: Google has integrated Gemini AI into Gmail's web and mobile interfaces, providing email summarization capabilities and advanced search features for subscribers. The feature is currently available only to Google One AI Premium ($20/month) subscribers and Google Workspace users Users can access Gemini through a distinctive black star-shaped logo in the Gmail interface The tool can summarize individual emails, threads, or specific time periods of correspondence Key features and operations: Gemini offers multiple ways to interact with and analyze email content, providing flexible summarization options across different timeframes. Users can request summaries through the Gemini button or specific commands like...
read Dec 2, 2024Ex Googler’s new site shows how much Google’s AI can glean from your photos
The growing concern over AI companies' access to and analysis of personal photos has led to the development of privacy-focused alternatives to mainstream cloud storage services. Privacy awakening: Former Google engineer Vishnu Mohandas left his position at the tech giant in 2020 after becoming increasingly concerned about how personal photos could be used to train AI systems. Mohandas developed Ente, an open-source photo storage service that features end-to-end encryption The platform aims to give users more control over their personal data while providing similar functionality to mainstream photo storage services Despite some limitations in features and ease of use compared...
read Dec 2, 2024ChatGPT is baffling users by refusing to say certain people’s names
ChatGPT has mysteriously stopped responding to prompts containing certain specific names, raising questions about content filtering and transparency in AI systems. Core issue identification: ChatGPT, OpenAI's popular language model, consistently returns error messages when asked to process or generate responses containing specific full names, including "David Mayer" and several others. Users across social media platforms have documented the AI's inability to combine certain first and last names, even though it can say the names individually The restriction appears to affect multiple versions of ChatGPT, including GPT-4 Other AI chatbots like Google Gemini and Grok have no difficulty processing these same...
read Dec 1, 2024Brands shift to AI-powered privacy-first advertising as consumer trust dwindles
The growing tension between AI-powered advertising and consumer privacy concerns is reshaping how brands connect with their audiences in the digital age. Current privacy landscape: Public perception of digital advertising has reached a concerning level of distrust, with 45% of UK consumers believing their phones are actively eavesdropping on their conversations to serve targeted ads. While phones don't actually need to listen in, the vast amount of behavioral data available to advertisers enables such precise targeting that it creates an illusion of surveillance The impending death of third-party cookies has pushed the advertising industry toward privacy-first solutions Growing ad clutter...
read Nov 27, 2024Bluesky blocks AI training on user posts, but can it stop others’ attempts?
The battle between social media platforms and AI data scrapers continues to escalate as Bluesky grapples with protecting user content from unauthorized AI training datasets. Recent incident sparks privacy concerns: A significant breach occurred when a Hugging Face employee scraped and published one million Bluesky posts to the AI repository, highlighting the vulnerability of public social media data. The dataset gained significant attention on the platform, trending throughout the day before being removed The employee has since apologized for the unauthorized data collection and removed the scraped content The incident exemplifies the ease with which public API data can be...
read Nov 27, 2024Microsoft denies using Office docs for AI training
Microsoft's privacy practices regarding AI training have come under scrutiny following recent misconceptions about how the company uses data from its Microsoft 365 suite of applications. Key misunderstanding clarified: Microsoft has explicitly stated that it does not use customer data from Microsoft 365 applications to train its large language models (LLMs). The controversy stemmed from confusion over a default privacy setting for "optional connected experiences" in Microsoft Office These connected experiences enable features like online picture search and internet-based information lookup The setting's disclosure language did not specifically address AI training, contributing to public confusion Technical context: The "optional connected...
read Nov 26, 2024Microsoft is training its AI on your Office docs — here’s how to stop it
Microsoft's use of Office documents for AI training has sparked concerns about data privacy and intellectual property rights, as users discover their content may be automatically included in AI model training without explicit consent. Key discovery: A cybersecurity expert from Cyberciti.biz has revealed that Microsoft's Connected Experiences feature automatically collects data from Word and Excel files for AI training purposes, with the feature enabled by default. The feature allows Microsoft to utilize various types of content, including articles, novels, and commercial works, for AI training This data collection occurs through Microsoft's Connected Experiences functionality within Office applications The company has...
read Nov 22, 2024Apple’s AI model will supercharge Siri but don’t expect it any time soon
The imminent transformation of Apple's Siri into a large language model (LLM)-powered assistant marks a significant shift in Apple's approach to artificial intelligence, though the implementation timeline stretches into 2026. Current state of Siri: Apple's intelligent assistant has lagged behind competitors in functionality and capabilities, often defaulting to web searches rather than providing direct answers. Despite Apple's claims of significant improvements in iOS 18, the promised "new era for Siri" has delivered only modest enhancements Third-party integrations remain limited compared to competitors like Amazon's Alexa The current version of Siri is widely considered less capable than Google Assistant and Alexa...
read Nov 20, 2024Strava limits third-party access to user fitness data
The fitness tracking platform Strava has announced significant changes to its API access policies, affecting how third-party applications can utilize user workout data. Key policy changes: Strava has implemented new restrictions on how third-party apps can access and display user activity data, particularly impacting training apps and AI-powered analysis tools. The platform, which serves over 100 million users, will no longer allow third-party apps to show Strava activity data to other users Applications are now prohibited from using Strava's API data in artificial intelligence models Third-party apps must complement rather than replicate Strava's interface and functionality Impact on training apps:...
read