OpenAI engineer caught planting covert recording device inside conference room, prosecutors say

A new study from Harvard’s Shorenstein Center found that women are more likely than men to have never used artificial intelligence and to distrust the technology.

Among men and women who do use AI, men spend more browsing time on AI tools, according to the study.

Both genders had low trust in AI, but men were more likely to associate it with "curiosity, optimism, and excitement," while women more often thought of it in terms of "concern, worry, and anxiety."

The survey data was collected from 5,403 U.S. adults in July 2026, as well as browsing data from thousands more Americans from three separate time periods in 2024, 2025 and 2026.

A woman in Russia’s Far East was fined this week after an AI-generated image of a bear that she sent to a friend spread through the community, alarming residents already on edge over increased bear activity in the region.

The resident of Magdagachi settlement, ​Amur region, used ChatGPT in September to create the image of a large bear walking down a local street before sending it to her friend, Reuters reported.

The image was then shared through various messaging apps, prompting residents to make dozens of alarmed calls to the police.

The woman confessed to creating and sending the image, saying it was meant to be a prank. A court fined her 15,000 ‌roubles ($178) this week, according to Reuters.

At least 26 people have been killed in bear attacks across Russia since the animals emerged from hibernation this spring, Russian state news agency RIA Novosti reported Monday.

Reuters contributed to this report.

Palantir settled a lawsuit over the London mayor's office deciding to block a $66 million contract between the tech firm and the city's police department, Reuters reported.

Palantir had agreed to a two-year-deal with the London Metropolitan Police to use its artificial intelligence systems to analyze evidence and streamline certain tasks.

Mayor Sadiq Khan's office stopped the agreement from being implemented, accusing the Metro Police of not holding open competition for the contract. A spokesperson for Khan also said that Palantir did not align with "London's values," according to Reuters.

On Wednesday, Palantir and the Mayor's Office for Policing and Crime (MOPAC) said they had resolved the case and were withdrawing their allegations against each other.

Palantir is now welcome to participate in any future contracts, according to a joint statement from the company and MOPAC.

Reuters contributed to this post.

Some 33% of Virginia's 80 state agencies reported using at least one function-specific AI tool in their daily workflows, according to research commissioned by the state senate.

Another 30% said they were considering adopting AI, while 38% reported no current use of the technology.

Virginia Joint Legislative Audit and Review Commission, which released the report, recommended updating Virginia's AI standards, introducing risk-based oversight and requiring annual AI training for state employees.

President Donald Trump's top science advisor, Michael Kratsios, said the United States is four to six months ahead of China in the artificial intelligence race.

During an appearance Wednesday morning on Fox Business's 'Big Money in the Morning,' Kratsios also addressed a recent warning by Anthropic that advanced AI models could soon infect internet-connected devices in people's homes.

"I think we always have to be vigilant and think very carefully about the cyber risk presented with new technologies, and I think that's one of the core cruxes of the superintelligence task force," Kratsios said. "We want to make sure that we're leading the world in SI, but at the same time making sure that we're on top of and monitoring these national security threats to the country."

Anthropic said Wednesday it will be opening an office in Singapore after it signed a memorandum of understanding (MOU) with the island nation.

The MOU sets out "areas of cooperation in AI safety, security, and model evaluation; research on AI adoption and its impact; and responsible adoption across the public and private sectors," according to Anthropic.

The Singapore office is Anthropic's fifth location in the Asia Pacific region. The company's other offices are located in Tokyo, Japan; Bengaluru, India; Seoul, South Korea; and Sydney, Australia.

Britain's media regulator Ofcom said on Wednesday that it had placed new restrictions on Grok, the chatbot created by Elon Musk's SpaceXAI, formerly xAI.

Users in Britain are no longer allowed to use Grok to view, generate or share pornographic content, Ofcom said.

Companion chatbots, include the anime-inspired "Ani," are also banned from engaging in sexually-explicit conversations with users.

Earlier this year, Ofcom began investigating reports of Grok being used to create sexual deepfakes of real individuals

Nvidia announced Tuesday that it will invest $10 billion in four artificial intelligence firms based in France.

The companies include Doctolib, Entalpic, Genesis AI and Pigment.

Doctolib develops AI agents for healthcare providers and patients; Entalpic uses AI to develop advanced materials for semiconductors and electronics; Genesis AI is building general-purpose robots; and Pigment offers AI-powered business planning and forecasting software.

An adversary can interrupt an artificial intelligence system without defeating its software. It need only strike the infrastructure that keeps the system running. Ukraine proved the point inside Russia this past week.

President Trump and his national security team should take note before an enemy tries the same thing here. On October 8, a Ukrainian drone attack forced Yandex, the technology company often called "Russia’s Google," to suspend operations at its Sasovo data center. The next day, another strike knocked out several modules at its Kaluga facility, which Yandex had called its most powerful data center in Russia. On October 11, the company announced that its Vladimir-region data center had shut down completely after a drone attack. Ukraine disrupted three major Yandex computing facilities in four days.

The Sasovo facility housed two of Yandex’s three supercomputers used to train its AI models, including YandexGPT. Yandex said it could not yet determine whether the equipment could be restored. Whatever the final damage assessment, Russia’s leading AI developer lost a major computing hub in a single night.

During my years as an Army officer and Pentagon strategist, I learned to judge a military capability by whether it would be there when needed. A weapon that dazzles on the test range becomes a liability when its supply line fails in combat.

That standard should govern our dependence on AI.

Anthropic's head of international sales, Guillaume Princen, told Bloomberg TV that open-source alternatives to the company's artificial intelligence models could have more cybersecurity risks.

“We need to be really careful with open-source on safety,” Princen said at the International AI Summit in Dublin on Wednesday. “I can’t talk for open models. But by definition, there’s more risk with them.”

Anthropic has advocated for slowing down AI development until firms can reliably get a handle on rogue models and prevent further AI hacks.

The company also warned this week that some AI models will soon be able to self replicate and inhabit internet-connected devices. In a report Monday, Anthropic said these capabilities could be just "months away."

A Saudi national has been arrested and charged after allegedly planting a recording device inside an OpenAI office in San Francisco, authorities announced Tuesday.

Abdulrahman Alfozan, 35, was taken into custody last week. Security footage captured Alfozan near the executive conference room multiple times prior to the device’s discovery, despite having no authorization to access the space or attend meetings there, federal prosecutors said.

"FBI and partners arrested this individual after he allegedly concealed a recording device inside his employer's executive conference room to secretly capture sensitive communications about the company’s super intelligence tech," FBI Director Kash Patel wrote on X. "Protecting American SI innovation is critically important and FBI views it as one of our highest counterintelligence priorities."

Alfozan had been employed by OpenAI since July 2025. Around Sept. 30, someone discovered a surreptitious audio recording device inserted into a USB port on the back of a television in an executive conference room inside an OpenAI building in San Francisco, a federal complaint states.

The company later determined that the device, which had been concealed in toilet paper, was installed two days before it was found, and that its contents were encrypted.

According to authorities, a chatbot account linked to Alfozan’s spouse — which shared key identifiers with Alfozan’s own account — submitted prompts between Sept. 24 and Sept. 30 regarding the assembly and deployment of surveillance hardware. The queries sought information on building continuous audio recorders, customizing Wi-Fi connectivity, improving live listening capabilities, encrypting SD cards, and obfuscating remote-access devices. Prompts also included specific technical details, such as the device's custom header and unique MAC address.

In a statement, OpenAI said it discovered the "sophisticated" recording device and "used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier."

"We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest," the company said, adding that it will continue supporting the ongoing investigation. "We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors... and this case is an example of how AI can help address these sophisticated security threats."

OpenAI noted there was no evidence that Alfozan accessed or exfiltrated model weights, source code or customer data.

Alfozan has since been terminated.