The Information AM

Save 25% on an annual subscription to read the most important news about technology and business first. For even more, Save $250 on The Information Pro for unlimited access to our proprietary org charts, databases and surveys.

An OpenAI employee, Abdulrahman Alfozan, has been arrested in connection with a hidden listening device uncovered at one of the company’s San Francisco offices, the Justice Department said on Tuesday. The employee, who has been terminated, was charged with “using or endeavoring to use an electronic, mechanical or other device to intercept oral communications.”

The company discovered the device late last month in an executive conference room, in which meetings were frequently held discussing confidential technical, commercial and strategic information, including research and product plans, the Justice Department said.

OpenAI used its own models to analyze the device and where it had come from, the company said on Tuesday. The company also said that it has no indication that Alfozan, who is a Saudi national, inappropriately accessed or exfiltrated model weights, source code or customer data.

Alfozan had worked at the company for about a year and was an engineering lead working on OpenAI’s agents product. He previously worked for five years at Meta and at an AI startup for logistics infrastructure.

This isn’t the first time a Saudi-connected individual was caught allegedly spying on a tech company. In 2022, a U.S. citizen who worked at Twitter was convicted of sending sensitive information about Twitter users to the Saudi government.

Apple is launching a tablet-like smart home device as part of a renewed push into the home.

The $299 HomeView is a 7-inch screen that sits on kitchen counters or mounts on walls. Apple said the HomeView is designed for watching and listening to content as well as smart home management. It can recognize up to six household members so it can display personalized information to them, such as calendar appointments and messages.

Apple’s recently overhauled Siri AI is the main interface for interacting with HomePod. While Apple used Google’s Gemini models to boost performance of a new version of Siri, some users have found the voice assistant experience remains lackluster compared to rivals at the cutting edge.

With HomeView, Apple is moving into a crowded space of smart home devices, where Amazon and Google have been competing for more than a decade. Apple has also launched home products before, with mixed success. Its HomePod smart speaker, first released in 2018, failed to catch on, due in part to Siri’s struggles to perform basic functions. Apple has also been selling an Apple TV set-top box since 2007. Apple unveiled a new Apple TV and HomePod mini on Tuesday.

Character.ai, a five-year-old startup that lets users create and interact with AI-powered personas, on Monday said it filed for Chapter 11 bankruptcy after encountering “significant legal, regulatory, and operational headwinds.” The startup said Disney, which will serve as the initial bidder in the bankruptcy auction process, will license some of Character.ai’s intellectual property for $20 million. Character.ai said it would continue to support subscribers to its platform.

The bankruptcy filing is the most severe outcome for an AI startup previously involved in what’s known as a reverse acqui-hire. In these arrangements, founders and key staff have left to work at a larger tech company, which agreed to license the startup’s technology. The fee typically compensates investors and employee shareholders. The remaining startups—including Inflection AI, whose founders joined Microsoft, and Adept, whose founders joined Amazon—have often instead have pivoted to other products.

But Character struggled with litigation tied to its chatbots and the costs of using external models after it stopped developing its own. In January, it settled a lawsuit with families alleging its chatbots had contributed to their children killing or otherwise harming themselves. Last year it considered selling itself or raising money, but did not take either action. In 2024, Google paid $2.7 billion to license Character.ai’s models and recruit co-founders Noam Shazeer and Daniel De Freitas back to Google. (Shazeer has since left Google for OpenAI.) Character’s remaining employees took ownership of the company.

Anthropic has struck an agreement to license technology and hire some engineers from Opaque Systems, a five-year-old startup that sells software for securing confidential data for use with AI agents, according to an Anthropic spokesperson.

The Anthropic spokesperson didn’t have a comment on the terms of the agreement or how many of Opaque’s engineers will be joining Anthropic. But Opaque CEO Aaron Fulkerson and co-founder and CTO Rishabh Poddar announced in LinkedIn posts that they will be moving over to Anthropic.

The agreement, which Opaque announced Tuesday, shows how Anthropic is trying to make it easier for customers in regulated industries like finance and healthcare to train AI models and run agents on confidential data, without exposing that data to unauthorized outsiders. Opaque, which began as a research project at UC Berkeley nearly a decade ago, lists Anthropic as a customer on its website, along with ServiceNow, Accenture and others.

The agreement is the latest example of how Anthropic has been building AI tools targeting specific industries including finance, cybersecurity and life sciences, which includes its acquisition earlier this year of AI biotech startup Coefficient Bio for around $400 million.

JPMorgan said it’s blocking external AI agents like Meta’s Muse and Instinct from accessing client accounts.

“We’re not going to allow just any outside agent to come in inside the company because that’s just too risky for clients and for JPMorgan,” said Jamie Dimon, CEO of the bank, on its third-quarter earnings call on Tuesday.

CFO Jeremy Barnum added, “to the extent that we’re able to block those, we are blocking them to protect customers and protect ourselves,” and customers will bear responsibility if they give up their credentials to the agent and the agent does something that they don’t want.

A JPMorgan spokesperson told CNBC that the bank doesn’t permit third parties to access accounts by obtaining customers’ banking credentials and logging in on their behalf. Earlier, Amazon also blocked Meta’s Muse from shopping on behalf of customers on the e-commerce site, citing privacy and security risks.

Data center operator AirTrunk is postponing a planned initial public offering of a real estate investment trust in Singapore to next year as higher interest rates and volatile markets dampen investor demand, Bloomberg reported.

The Sydney-headquartered company, which is backed by Blackstone, had plans to start gauging investor interest for a $1.5 billion IPO in October. AirTrunk had explored offering a yield of about 6% for the REIT to attract investors, but was put off by September’s U.S. interest rate hike and the prospect of further increases. Dow Jones Newswires first reported on the IPO delay.

AirTrunk is the latest AI-linked company to pull back from the public markets. Last week, Nvidia-backed AI cloud company Firmus scrapped what would have been Australia’s biggest initial public offering in nearly 30 years. Investors balked at the company’s proposed valuation of $30.6 billion, which looked too high given a thin track record of operating data centers and a significant debt pile.

The data center arm of Philippine telecom company PLDT said last week it would postpone its REIT listing–set to be the country’s first REIT IPO–until next year amid rising interest rates.

Microsoft has discussed making significant cost reductions in its Azure cloud unit, The Information reported late Tuesday. Girish Bablani, president of Microsoft’s Azure cloud unit, recently told senior leaders in his group to think about ideas for reducing costs by at least 20%, according to a person with direct knowledge of the situation.

The talks at Microsoft, which come amid a major reorganization of Azure announced internally earlier this week, illustrate how the software giant is dealing with challenges such as the AI talent wars. Rivals like Anthropic and Meta have been poaching some of its top employees, and in some cases Microsoft has had to spend more to counter those offers, as we reported.

Microsoft, like other major cloud providers, has also cut costs through layoffs after rising capital expenditures for AI data centers reduced their respective free cash flows.

JPMorgan CFO Jeremy Barnum said on Tuesday that the bank is carefully monitoring its AI exposure as the capital expenditure demands of the AI buildout grow. He noted that lenders generally need to keep tabs on direct and indirect exposure to AI infrastructure spending, as well as other factors like the value of collateral.

The comments, on JPMorgan’s third-quarter earnings call, offered a view into how the biggest U.S. bank is managing AI-related risk, with Barnum noting that AI-related financing is increasingly pulling from a variety of markets. “It’s not just traditional construction lending. Now increasingly, it’s working capital revolvers. You also see the equity markets, it’s kind of everywhere,” he told analysts.

More broadly, Barnum said that the AI buildout increasingly requires financing from different parts of the capital markets ecosystem, including from banks as well as non-bank lenders. He noted that while JPMorgan’s AI exposures are still “quite modest” overall, the bank could still feel an impact from a major sentiment shift in AI. “I would be surprised if anyone, frankly, was immune at this point,” he said.

Barnum’s comments land as Wall Street has grown more weary of the buildout in recent weeks, with some banks stepping back from data center loan deals and delays for several AI-related IPOs.

This week on AI Deep Dive

Check out the most recent episode of AI Deep Dive. A reported