Congressional Democrats on the Joint Financial Committee launched a report this week pinpointing greater than $20.9 billion in shopper losses stemming from identification theft that got here out of 4 main breaches of knowledge dealer companies. US senator Maggie Hassan launched the investigation in August after an investigation by The Markup and CalMatters, copublished by WIRED, discovered that some information brokers had been hiding opt-out instruments from Google and different search engines like google and yahoo.
The US Division of Justice’s current launch of three million paperwork associated to convicted intercourse offender Jeffrey Epstein included grand jury subpoenas to Google that make clear how federal investigators work together with tech corporations and the way they reply to authorities requests for data.
The Mexican drug cartel CJNG might survive the killing of its longtime chief Nemesio “El Mencho” Oseguera Cervantes partially because of its prolific use of applied sciences like drones, social media, and AI. In the meantime, the Mexican Navy introduced on Thursday that it had seized a semi-submersible vessel carrying practically 4 tons of cocaine as a part of a current initiative to discourage drug trafficking within the Pacific Ocean. The trouble comes because the US has launched its personal purported marketing campaign in opposition to maritime trafficking through a collection of lethal assaults on boats within the Caribbean.
In the meantime, as AI assistant brokers like OpenClaw explode in recognition—and sow chaos across the internet—a brand new open supply undertaking known as IronCurtain is utilizing a singular design to safe and constrain agentic AI earlier than it may possibly go rogue.
And there’s extra. Every week, we spherical up the safety and privateness information we didn’t cowl in depth ourselves. Click on the headlines to learn the complete tales. And keep secure on the market.
Setting an autonomous internet-enabled robotic free in your home ought to give anybody a second’s pause. When that robotic is a roving vacuum cleaner geared up with a digital camera and microphone that may very well be hijacked from wherever on the planet with nothing greater than its serial quantity, it turns into an precise privateness horror story.
One such robovac proprietor, Sammy Azdoufal, found that absurd safety vulnerability whereas trying an experiment in piloting his DJI Romo robotic vacuum cleaner with a PS5 controller. He discovered that he might as an alternative management 6,700 of the robots in 24 nations world wide, with full entry to the ground plans they generated of their homeowners’ houses and their video and audio feeds. When The Verge contacted Azdoufal, he was in a position to immediately entry a Romo owned by a staffer on the tech information outlet simply by figuring out its 14-digit serial quantity. DJI has now fastened the vulnerability in response to Azdoufal primarily live-tweeting his findings. However the story nonetheless raises severe questions concerning the safety of different audio- or video-enabled internet-of-things devices—to not point out ones able to freely roaming your house.
Whereas the Division of Homeland has been massively empowered beneath the Trump administration in its mission to deport tens of millions of immigrants, the group inside DHS that serves as the USA’ major cyber defender, the Cybersecurity and Infrastructure Safety Company, has been uncared for. Now its performing director, Madhu Gottumukkala, has been changed as CISA seeks to seek out new footing.
Even earlier than that information, CyberScoop this week reported on the crises which have plagued the company for the complete first yr since Trump’s inauguration: A 3rd of the workers has been laid off and full divisions of the company have been closed. Nominations for a everlasting director have been blocked in Congress. Its capabilities have withered, and organizations that had sought out CISA for help and partnerships have regarded elsewhere. Gottumukkala has suffered his personal extra private scandals reminiscent of ousting safety personnel after he failed a polygraph check and sharing delicate contracts on ChatGPT. Now Nick Andersen, who has served as CISA’s govt director for cybersecurity, will exchange Gottumukkala on the beleaguered company.
A researcher at King’s Faculty London pitted three well-liked massive language fashions in opposition to one another in simulated conflict recreation eventualities and located that, 95 % of the time, not less than one of many fashions opted to deploy tactical nuclear weapons. The researcher additionally discovered, when an AI mannequin deployed a tactical nuclear weapon, its AI opponent solely deescalated a fourth of the time. Not one of the corporations behind the three fashions—OpenAI, Google, and Anthropic—responded to New Scientist’s request for remark.
AI’s function in war-fighting has lurched into the highlight this week. Anthropic and the Division of Conflict are embroiled in a contract dispute over whether or not Anthropic’s AI fashions can be utilized to energy totally autonomous weapons and mass home surveillance. Dario Amodei, Anthropic’s CEO, wrote in a assertion that all these use instances “can undermine, slightly than defend, democratic values.” In flip, President Donald Trump has threatened to ban the usage of Anthropic merchandise, together with its Claude chatbot, throughout the US authorities. In the meantime, tons of of Google and OpenAI staff have signed an open letter asking for his or her bosses to “put apart their variations and stand collectively to proceed to refuse the Division of Conflict’s present calls for for permission to make use of our fashions for home mass surveillance and autonomously killing individuals with out human oversight.”
A brand new app for Android telephones known as Close by Glasses lets customers scan for good glasses in your neighborhood, revealing the presence of the wearable devices, that are typically indistinguishable from regular glasses and let wearers report individuals with out their data. The app scans for the distinctive Bluetooth signatures that the glasses emit, and sends customers a notification if it detects a close-by supply.
The developer informed 404 Media that he was impressed to construct the app after studying about a number of incidents involving good glasses. Over the summer time, 404 Media reported {that a} Customs and Border Safety agent had donned a pair throughout an immigration raid, and this fall the outlet additionally reported that males had been utilizing good glasses to movie therapeutic massage parlor staff, seemingly with out their data or consent. In February, The New York Instances reported that one smart-glasses developer, Meta, had plans to combine face recognition into its glasses, spurring recent considerations amongst privateness specialists.
