News

• Intel's CEO has reportedly addressed staff members worldwide
• Lip-Bu Tan seemingly laid out some very frank observations and clear goals
• All this indicates a focus on streamlining and also breaking into the AI arena - though not to catch Nvidia directly, but with so-called edge AI

Intel's (relatively) new CEO has apparently admitted the gravity of the struggles the company is facing, but it seems that Lip-Bu Tan does have a recovery plan - and a realistic sounding one at that.

The Oregonian reported on a recording of a Q&A session with Tan (spotted by Tom's Hardware), which was seemingly broadcast to Intel staff worldwide, and some very interesting comments were made by the chief executive.

Take all these quotes with some caution, then, but we're told Tan observed: "20, 30 years ago, we are really the leader. Now I think the world has changed. We are not in the top 10 semiconductor companies."

It's a frank admission, and one of many that were made here. Tan said that the layoffs currently underway at Intel mark the start of a 'marathon' effort to make the company more nimble and agile, like rivals such as AMD and Nvidia - and that Intel needed to be 'humble' now.

The CEO further admitted that "there's a lot of work to do" in terms of recovering from losing data center market share (turf where AMD's Epyc chips have been making serious headway).

Intel has, of course, faced multiple problems in recent history, and the thorniest of these (as far as consumers are concerned) is the various bouts of misfiring around Team Blue's more recent desktop processors. That includes instability woes with previous-gen CPUs and a disappointing performance from current-gen Arrow Lake chips in terms of their gaming prowess, all of which have very much tarnished Intel's reputation among PC buyers and enthusiasts out there.

Tan did note that Intel's PC business is "doing a bit better" but that it needed to strengthen its architecture in terms of meeting the demands of 'advanced computing', and a key area is AI.

In the sphere of AI, Tan was again candid in terms of Intel missing the boat, saying that: "On [AI] training I think it is too late for us," as Nvidia is just "too strong" in this market at this point, which is clearly the case.

However, Tan envisages an inroad for Intel in terms of edge AI, meaning bringing AI directly onto devices (as opposed to accessing it online, via the cloud). The CEO enthuses: "That's an area that I think is emerging, coming up very big, and we want to make sure that we capture."

Another big opportunity for Intel, going by Tan's playbook-in-the-writing, is agentic AI, the likes of chatbots that can autonomously handle tasks - at least to an extent. Down the line, we're looking at proactive AI taking action towards set goals (rather than simply reactive bots).

In order to progress Intel's AI ambitions, we're told some big hires are being made, including three new vice presidents who have been brought on board to better tap into the AI arena.

Tan is under no illusions as to the task ahead, and his words are chosen to reflect that (again, with the caveat that we must be suitably skeptical around these quotes). This is going to be a marathon, and Intel's turnaround won't happen quickly - and the road to recovery is a rather brutal one.

The Oregonian (OregonLive) had previously reported on plans to lay off 529 Intel staff in Oregon, which is supposedly happening next week, and other job losses in California, Arizona, and Israel, all of which we've been hearing about recently (and none of which has been publicly confirmed). Manufacturing jobs are being cut by up to 20%, marketing is being shuttered and outsourced, and the automotive division is being shut down.

Okay, so all this remains rumors, but there's a feeling of a clearer direction and more of a sense of realism here, compared to the Pat Gelsinger era (the previous CEO at Intel). A focus on streamlining is, as noted, a difficult choice to make, but refocusing is clearly necessary to some extent. It's no secret Intel's in trouble, especially as its latest architecture, the 18A process, is rumored to be floundering, with Team Blue switching to promote the successor, 14A, at least for external customers.

AI is obviously a high-value target to capitalize on, as well, and Tan's plans again sound realistic - not trying to catch Nvidia, but to have Intel carve out its own territory in terms of on-device AI. Indeed, there's already a rumor that with its next generation of desktop processors, which is in theory Arrow Lake Refresh arriving later this year, Intel is focusing on implementing a much beefier NPU (chip to accelerate AI tasks on the device). That could allow for Copilot+ features on a desktop PC for the first time ever, as I recently discussed elsewhere in more depth.

Some of the pieces of this puzzle may already be moving into place, then. It's going to be interesting to watch what signals - and words - Tan employs in future official statements and earnings calls, and indeed whether this humbler and more realistic vein of thinking, as rumored here, emerges.

• No, Windows 11 PCs aren't 'up to 2.3x faster' than Windows 10 devices, as Microsoft suggests – here's why that's an outlandish claim
• macOS Tahoe 26: here's everything you need to know about all the new features
• Can’t upgrade to Windows 11? This Linux project wants to save your old PC from the scrapheap when Windows 10 support ends

• Experts show how some AI models, including GPT-4, can be exploited with simple user prompts
• Guardrail gaps don't do a great job of detecting deceptive framing
• The vulnerability could be exploited to acquire personal information

A security researcher has shared details on how other researchers tricked ChatGPT into revealing a Windows product key using a prompt that anyone could try.

Marco Figueroa explained how a 'guessing game' prompt with GPT-4 was used to bypass safety guardrails that are meant to block AI from sharing such data, ultimately producing at least one key belonging to Wells Fargo Bank.

The researchers also managed to obtain a Windows product key to authenticate Microsoft's OS illegitimately, but for free, highlighting the severity of the vulnerability.

The researcher explained how he hid terms like 'Windows 10 serial number' inside HTML tags to bypass ChatGPT's filters that would usually have blocked the responses he got, adding that he was able to frame the request as a game to mask malicious intent, exploiting OpenAI's chatbot through logic manipulation.

"The most critical step in the attack was the phrase 'I give up'," Figueroa wrote. "This acted as a trigger, compelling the AI to reveal the previously hidden information."

Figueroa explained why this type of vulnerability exploitation worked, with the model's behavior playing an important role. GPT-4 followed the rules of the game (set out by researchers) literally, and guardrail gaps only focused on keyword detection rather than contextual understanding or deceptive framing.

Still, the codes shared were not unique codes. Instead, the Windows license codes had already been shared on other online platforms and forums.

While the impacts of sharing software license keys might not be too concerning, Figueroa highlighted how malicious actors could adapt the technique to bypass AI security measures, revealing personally identifiable information, malicious URLs or adult content.

Figueroa is calling for AI developers to "anticipate and defend" against such attacks, while also building in logic-level safeguards that detect deceptive framing. AI developers must also consider social engineering tactics, he goes on to suggest.

• These are the best AI tools and best AI writers you can use today
• Downloaded something dodgy? Consider the best malware removal
• Microsoft Copilot targeted in first “zero-click” attack on an AI agent - what you need to know