🦆 DuckDuckGo's 'No AI' search triples in traffic

💸 GitHub Copilot switches to token-based billing

⚖️ Microsoft threatens researcher over exploit disclosure

🖱️ Meta employee tracking may breach EU rules

👓 Apple AI glasses delayed to late 2027

+ 📊 Daily poll and results

+ 📈 Trending tools and resources

DuckDuckGo says traffic to its opt-in "No AI" search page has tripled since Google pushed its latest AI search overhaul, and the numbers are still climbing as users seek an escape route.

Google recently moved AI mode into a more central spot on its homepage, giving it priority over the classic ten links. CEO Sundar Pichai has loudly promoted AI as core to Google's future.

DuckDuckGo isn't anti-AI here. It also runs duck.ai for the maximalists, happily serving both crowds. The company plays both sides of the aisle and lets users pick their poison.

GitHub Copilot moves from a flat subscription to token-based billing on June 1, charging users by how many tokens they burn instead of a low fixed rate tied to requests.

Developers flooded Reddit and X with complaints, with one claiming costs jumped from $29 to nearly $750 a month and another sharing a screenshot showing a leap from $50 to roughly $3,000.

Critics of the complainers blame "vibe coders" who burn tokens carelessly, but others point the finger at Microsoft, which spent months encouraging users to lean on Copilot exactly this hard.

Microsoft threatened a criminal case against a researcher called Nightmare Eclipse, who publicly posted proof-of-concept zero-day exploit code, then disabled their GitHub, GitLab, and Microsoft Security Response Center accounts.

Microsoft says Nightmare Eclipse skipped "proper coordination" when reporting the vulnerabilities. Posts suggest the person is a disgruntled former employee, which security researcher Kevin Beaumont flagged after reviewing the company's public response.

Beaumont notes Microsoft itself hires people who posted zero-days, some carrying hacking convictions, and even buys exploits from brokers. Banning someone also makes "responsible" future reporting tricky, as he dryly pointed out.

Reuters reports that Meta's employee tracking tool, the Model Capability Initiative, captures keystrokes, mouse movements, and clicks across 200-plus apps to train AI, and may sweep up non-US data too.

Meta admits the tool grabs emails and chats US staff send abroad, no matter where the other person sits. A legal expert told Reuters this could violate the EU's GDPR, which demands a legal basis for collecting personal data.

Meta employees have protested since launch, distributing flyers and a petition. Some worry they're training their own replacements, while others burn through monthly data quotas in days thanks to MCI's appetite.

Bloomberg's Mark Gurman reports Apple pushed its N50 smart glasses, a hardware platform for Siri and Visual Intelligence, from a late-2026 reveal to a launch by the end of 2027.

Apple worried its visual AI technology wouldn't be ready and didn't want to ship an unappealing product. The first glasses will feature oval cameras, unique colors, and multiple frame styles, later expanding into health and AR.

Gurman also says the lighter Vision Air may arrive in 2028 or 2029, calling the headset category "on ice" until then. So the current Vision Pro stands alone for another two or three years.

A) Yes - 31%

B) No - 69% 🏆