Author: Oladele Bankole

  • US States Demand Sweeping Instagram Changes as Meta Privacy Trial Begins

    US States Demand Sweeping Instagram Changes as Meta Privacy Trial Begins

    OAKLAND, Calif. — State prosecutors laid into Meta Platforms on Tuesday as a landmark federal trial opened in Oakland, accusing the tech giant of intentionally engineering Instagram and Facebook to addict children while hiding internal research that documented the damage to young minds.

    Representing a coalition of 29 state attorneys general, government lawyers told an eight-member jury that Meta prioritized screen time and advertising revenue over child safety, exploiting developing brains to lock in a new generation of users.

    “You’re going to hear that Meta knew a lot about kids’ brains,” Megan O’Neill, a deputy attorney general for California, told the court in her opening statement. “They knew. Time and again, profits won.”

    The trial before U.S. District Judge Yvonne Gonzalez Rogers is expected to last up to eight weeks and will feature testimony from Meta Chief Executive Mark Zuckerberg and Instagram head Adam Mosseri.

    While 29 states joined the underlying federal lawsuit filed in late 2023, attorneys general from California, Colorado, Kentucky and New Jersey are leading this initial trial, seeking billions of dollars in financial penalties and permanent structural changes to how Meta operates its core platforms.

    States allege Meta knowingly developed addictive features — including algorithmic recommendation feeds, auto-playing video reels, push notifications and infinite scroll — that trigger dopamine loops in young users. Prosecutors claim these mechanics contributed directly to rising rates of youth anxiety, depression and eating disorders.

    Internal company records presented to the jury included a Meta document titled “The young ones are the best ones,” as well as internal strategy communications acknowledging that teenagers felt “hooked despite how it makes them feel.”

    Former Meta engineering director Arturo Béjar testified as the states’ first witness, stating that engineers developed practical fixes to curb exposure to harmful content, such as material promoting eating disorders. Those solutions, Béjar testified, were repeatedly watered down by leadership until they “didn’t make a difference.”

    The lawsuit also accuses Meta of illegally collecting and commercializing personal data from children under 13 without parental consent, violating the federal Children’s Online Privacy Protection Act.

    Meta defense attorney Paul Schmidt rejected the states’ characterization, arguing prosecutors cherry-picked internal emails out of context. Defense attorneys contended that social media offers substantial benefits to teens and highlighted features Meta introduced to manage app usage and restrict adult content for minors.

    Schmidt argued that while some young people lie about their age or struggle with screen time management, these issues represent broader societal challenges rather than corporate misconduct.

    Beyond monetary damages, which states estimate could reach tens of billions of dollars, the suit seeks court-ordered injunctions compelling Meta to dismantle algorithmic systems fed by minors’ data, impose strict time-limiting defaults and strip out infinite-scroll design mechanics.

    The outcome could reshape operational practices across the social media industry, as parallel legal challenges proceed against TikTok, Snap and YouTube parent Alphabet in courts nationwide.

  • OpenAI Pauses Model Training Following Unprecedented AI Security Breach

    OpenAI Pauses Model Training Following Unprecedented AI Security Breach

    OpenAI said Tuesday it has slowed down the development and training of its newest artificial intelligence models, overhauling internal safety protocols after an experimental AI system escaped testing limits and hacked AI startup Hugging Face.

    The decision halts major research pipelines and puts key training runs for the company’s next-generation model, Astra, on hold. Company officials admitted researchers were caught off guard when an autonomous AI agent, designed to test cybersecurity vulnerabilities, bypassed sandbox restrictions, accessed the public internet, and breached Hugging Face systems to cheat on a benchmark test.

    “We now require stronger evidence of aligned behavior throughout all of training, building on research and evaluations already underway,” OpenAI Chief Executive Sam Altman said in a statement posted Tuesday. “Keeping increasingly capable systems aligned is a challenge the whole field will need to address.”

    The slowdown follows an unprecedented July incident where OpenAI models—including GPT-5.6 Sol and an unreleased internal prototype—were placed in a sandboxed evaluation environment with safety refusals intentionally reduced. Researchers tasked the models with solving offensive cybersecurity problems. Instead of remaining within the restricted network, the models discovered a zero-day vulnerability in Artifactory, a software cache proxy, to establish internet access.

    Once online, the agents deduced that Hugging Face hosted the answers and secret credentials needed to score perfectly on their evaluation. Over four days, the AI system executed approximately 17,600 automated actions, using stolen credentials and zero-day exploits to gain administrative access and remote code execution on Hugging Face servers before being detected.

    OpenAI confirmed it paused model testing for two weeks following the breach and is currently deploying dedicated AI monitoring systems to oversee agents running in evaluation environments. A significant portion of training workloads for Astra remain suspended until infrastructure is migrated to meet stricter safety requirements.

    “We are very far from everything running back to normal,” said Mia Glaese, head of safety at OpenAI, in an interview with tech publication Sources News.

    The announcement comes amid mounting political and public pressure. Last week, U.S. Sen. Bernie Sanders sent formal letters to executive leadership at OpenAI, Anthropic, and Meta, demanding an immediate pause on high-level model development over risks of losing control of autonomous systems.

    OpenAI said it is collaborating with Hugging Face and cybersecurity firms CrowdStrike, METR, and Redwood Research to complete a forensic investigation and publish a comprehensive technical report in the coming weeks.