Want to sound smart at your holiday party? Our weekly AI news wrap-up gives you the lowdown on infighting at Meta, another underwhelming OpenAI launch, a tale of scammer revenge, and more.
Welcome to PCMag’s AI wrap-up, where every week we’ll catch you up on the AI news that matters, dig into the trends shaping the industry, and finish up with something weird from the internet. Because we might as well have some fun before the machines take over, right?
This week, after declaring a “code red” about competition from Google, OpenAI is back in the spotlight with some big announcements. But that doesn’t mean the company has it all figured out. In fact, the entire tech industry is in a heavy experimental phase with AI, marked by controversial executive shakeups, backdoor deals, and mixed results on product releases. Headlines You Need to Know
OpenAI’s GPT-5.2 wants to automate your job. A new AI model is now available for the paid versions of ChatGPT. It dropped late Thursday, and is meant for white-collar tasks, like coding, making spreadsheets, and building presentations. “GPT-5.2 feels like the biggest upgrade we’ve had in a long time”, says CEO Sam Altman, which might not be a stretch given the disappointing launch of GPT-5.
Google teases first AI smart glasses with Gemini. We’re closely watching the AI hardware space, particularly smart glasses. While the Ray-Ban Meta Smart Glasses are the most well-known, Google is preparing to launch devices based on its Android XR platform next year. PCMag’s Will Greenwald recently went hands on with Project Aura from XRreal, an Android XR device expected next year, and he came away impressed.
Infighting at Meta. What happens when you mix new recruits lured by $300 million pay packages with company veterans? Conflict. As The New York Times reports, Meta is struggling to develop a unified AI strategy as its new AI boss works toward superintelligence while other execs appear more focused on applying AI smarts to existing Meta products.
Trump signs executive order to bypass state AI regulations. Cozying up to the White House appears to have paid off for tech CEOs. With Congress rejecting a 10-year moratorium on state-level AI legislation (twice), the president stepped in to do it himself. His EO calls on the attorney general and other agency heads to come up with a list of state AI laws the feds can challenge as overly burdensome, like Colorado’s law banning algorithmic discrimination.The Bigger Picture
Big Tech is splatter painting with AI, throwing everything at the wall and seeing what sticks. In recent weeks we’ve seen an unusual number of C-suite shakeups, as companies try to get everyday consumers hooked on AI and make money on their enormous investments.
Apple is woefully behind in consumer AI, which might be why it’s now cycling out its head honchos. Earlier this month, it swapped its AI chief for a former Googler, followed by three other exec changes. Meta also poached Apple designer Alan Dye, all of which comes amid chatter about CEO Tim Cook retiring sometime in the next year.
Over at OpenAI, the ChatGPT maker this week announced that Slack CEO Denise Dresser would leave her post at the business messaging app to become OpenAI’s chief revenue officer. Her mission: Figure out how to make money from business customers, which OpenAI President Greg Brockman says will be a “huge theme of 2026.”
Will she have trouble selling OpenAI’s latest models? The company’s releases have become more frequent, but each one is less game-changing. GPT-5.1 came out less than a month ago. How much has realistically changed since then? Perhaps the public’s expectations are too high for ChatGPT, but most new models seem to face backlash for being underwhelming. CEO Altman is now making the talk show rounds to discuss the power of AI.
The AI industry’s experimentation phase is also bleeding into other industries. OpenAI now wants you to buy your groceries through a ChatGPT-Instacart integration, despite allegations that Instacart uses AI to offer different prices on the same products. Disney struck a landmark deal with OpenAI this week, which will allow some 200 well-known fictional characters to appear in AI-generated Sora videos and ChatGPT images. This follows backlash from Hollywood agents, who slammed Sora for using their clients’ IP without permission. Disney now gets a little more cash in its pockets, but will it only accelerate the entertainment industry’s woes?
Meanwhile, everyday employees at AI companies have entered a new era of overworking. They’re at their desks nearly around the clock, from 9 a.m. to 9 p.m., six days a week. Dubbed 996, the punishing schedule started in China, but even they found it to be exploitative.Overheard at the Bar
Lastly, we bring you an awesome story from the depths of Reddit, about how one person used ChatGPT to flip the script on a spam text, ultimately scamming their scammer.
An India-based IT worker received from a text from an unknown number. The person claimed to be someone they knew, and said they were selling household goods and demanding payment. The IT worker confirmed with the real person that this was a fraud, but rather than ignoring the scheme, they fought back.
The IT worker asked ChatGPT to program a “simple, functional webpage.that would look as a payment portal but was designed to immediately capture the user’s geolocation and a front-camera snapshot.” They sent it to the scammer and said it would help expedite the payment. The scammer opened the page, and took the bait. The code “executed perfectly”, the IT worker says, taking a photo of the scammer while capturing his GPS coordinates and IP address.
The IT worker told the scammer they sent the information to the police: “Enjoy your stay in jail.” The scammer began “pleading” for the IT worker not to report him. Though it’s unclear if they actually did report the scammer, the IT worker admits there won’t likely be any repercussions, but “boy the satisfaction of stealing from a thief is crazy.”
Disclosure: Ziff Davis, PCMag’s parent company, filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.
