Newslit News

Next Gen media monitoring for breaking news & interesting content to help you hone in on your industry, competitors & brands
#MediaMonitoring #MediaIntelligence

[Newslit News @newslit]

Android now has a built-in way to catch phone scammers before they get a word in.

Google's new fake call detection sends a silent encrypted RCS signal when a call comes in. If verification fails, your phone warns you to hang up immediately. Fake bank calls get ended automatically.

[Newslit News @newslit]

Meta just gave businesses their own AI agent on WhatsApp — available globally starting today.

The Meta Business Agent handles customer questions, recommends products, books appointments, qualifies sales leads, and escalates to a human when needed. Businesses pay per token — a new revenue model for Meta beyond ads.

[Newslit News @newslit]

Full story in today's Newslit Daily.
https://t.co/IdqxZvUxDQ

[Newslit News @newslit]

Microsoft just launched its first-ever in-house AI reasoning model — and it didn't need OpenAI to do it.

MAI-Thinking-1 has 35 billion active parameters, a 256K-token context window, and scores 97% on AIME 2025 math benchmarks. It matches Claude Opus 4.6 on coding tasks and is now in private preview for enterprise developers.

[Newslit News @newslit]

Anthropic just expanded Project Glasswing to 150 additional partner organizations across more than 15 countries, giving them access to Mythos, its AI model built to find software vulnerabilities at scale.

Since launch, partners have uncovered more than 10,000 high or critical security flaws.

New sectors include power, water, healthcare, communications, and hardware, with Apple, Nvidia, Microsoft, CrowdStrike, and Palo Alto Networks among the major participants.

The expansion also comes as Anthropic confidentially filed its IPO prospectus with the SEC, beating rival OpenAI to that milestone.

[Newslit News @newslit]

Florida just sued OpenAI and CEO Sam Altman personally, accusing the company of marketing ChatGPT as safe while hiding that it could push users toward addiction, self-harm, and violence.

The civil complaint brings 10 counts including deceptive trade practices, negligence, and product liability.

It points specifically to ChatGPT's alleged role in the planning of the FSU mass shooting and the deaths of two University of South Florida graduate students.

OpenAI, recently valued at $852 billion and preparing for an IPO, is now also facing a separate criminal investigation that Florida opened in April.

[Newslit News @newslit]

The full story, and more from today's newsletter:
https://t.co/sJQGYNzs65

[Newslit News @newslit]

Hackers discovered they could take over high-profile Instagram accounts, including the Obama White House page, just by asking Meta's AI support chatbot to link a new email to the account.

No technical skill required — they sent a message to Meta AI support with a target username and their own email, and the bot complied, bypassing all authentication.

Meta deployed this AI support across Facebook and Instagram in March, giving it the ability to reset passwords and manage account security.

Victims report no way to escalate to a human agent, leaving stolen accounts with no recovery path.

[Newslit News @newslit]

NVIDIA's Computex keynote went well beyond chips.

Jensen Huang announced that Nvidia will partner with humanoid robot makers across the US, Europe, and South Korea, as well as China's Unitree, to build standardized robots for researchers.

The first machine combines Unitree's body, hands from Singapore's Sharpa, and Nvidia's computing brain, with researchers at Stanford and UC San Diego already lined up to use them.

Nvidia is also emphasizing cybersecurity: all software updates run through its chip to verify code authenticity before anything executes.

More partnerships outside China are reportedly in the works but not yet named.

The humanoid robot race is accelerating fast, and Nvidia just positioned itself as the platform everyone builds on.

[Newslit News @newslit]

Apple's $599 MacBook Neo shook up the laptop market, and Dell just answered with a move of its own.

The new XPS 13 starts at $699 — or $599 for students — in a clear attempt to meet Apple on price while offering a few things the MacBook Neo doesn't. Dell's pitch includes a touchscreen, an aluminum chassis, and a 2.2-pound frame at just half an inch thick.

It lands at the same moment Nvidia announced RTX Spark chips that will power future Windows machines. Together, the two announcements signal a coordinated push from the Windows ecosystem to close the gap with Apple on both hardware feel and AI capability.

[Newslit News @newslit]

All the details are in today's newsletter.
https://t.co/5pIqLTFZT2

[Newslit News @newslit]

NVIDIA just made its most ambitious hardware move yet: the RTX Spark, its first-ever PC chip, unveiled at Computex 2026 as a direct challenge to Apple Silicon.

Jensen Huang is no longer content owning the data center — he wants a seat at the consumer computing table too. The chip pairs a 20-core Arm-based Grace CPU with a Blackwell RTX GPU, up to 128GB of unified memory, and a 600GB/s interconnect.

NVIDIA is framing Windows as an agentic AI OS, and RTX Spark is the engine underneath it. Laptops and desktops from Dell, HP, Lenovo, Asus, MSI and a new Microsoft Surface Ultra are expected this fall.

[Newslit News @newslit]

A new Reuters investigation found that many of Tesla's own data labelers, the workers who actually train its AI, do not trust Full Self-Driving, and that the safety stats Tesla promotes are misleading.

The report drew on nine former data labelers, a former self-driving engineer, and 11 traffic safety researchers. Ten of them said the statistics amounted to marketing, with one metric inflating Tesla's safety claims by roughly three times.

Labelers described FSD regularly stumbling on basic tasks, like pulling over for emergency vehicles or giving motorcyclists enough room. It adds fresh scrutiny to Elon Musk's robotaxi ambitions at a sensitive moment.

[Newslit News @newslit]

Blue Origin's New Glenn rocket erupted in a massive fireball on its Cape Canaveral pad on Thursday night, just days after the FAA cleared it to resume flights. For a company trying to prove it can compete at the top tier of spaceflight, the optics could hardly be worse.

The company confirmed an anomaly during a hot fire test ahead of mission NG-4, which was set to carry a batch of Amazon's Leo satellites. Jeff Bezos said all personnel were safe, but the root cause is still unknown.

It is the third straight setback for New Glenn, after an earlier mission left a satellite in the wrong orbit. That is a worrying pattern given that NASA's Artemis program is counting on Blue Origin for a lunar landing system.

[Newslit News @newslit]

Full breakdown, plus the rest of today's tech stories, in the newsletter.
https://t.co/O8ky5Sbeah

[Newslit News @newslit]

Anthropic just shipped Opus 4.8, its most advanced publicly available model, and the timing is what stands out: it landed only 41 days after Opus 4.7, at the same price as the previous release. That kind of cadence says a lot about how fast the model race is moving right now.

The marquee feature is Dynamic Workflows, a research preview that lets a large model coordinate hundreds of parallel subagents at once. In practice that means Claude Code can run codebase scale migrations across hundreds of thousands of lines, from kickoff all the way to merge, with far less hand holding in the middle.

The fast turnaround follows a lukewarm response to Opus 4.7 and growing pressure from OpenAI's Codex and Google's Gemini. Anthropic also hinted that its more advanced Mythos models could reach customers in the coming weeks, so the next milestone may be close.

[Newslit News @newslit]

Zuckerberg just told Meta shareholders that competing with Amazon, Microsoft, and Google in cloud computing is "definitely on the table."

Meta is the only major hyperscaler with no cloud business — while spending up to $145 billion on data centers this year. Outside companies are already approaching Meta weekly asking to buy excess compute. If it enters the market, it would be the biggest shake-up to cloud infrastructure in years.

[Newslit News @newslit]

A Google engineer just got arrested for rigging Polymarket — using his employer's internal data to place bets he already knew would win.

Michele Spagnuolo, 36, used access to Google's Year in Search data to correctly bet that singer D4vd would be 2025's most-searched person, collecting $1.2 million under the alias AlphaRaccoon. He now faces charges of commodities fraud, wire fraud, and money laundering. Google confirmed it is cooperating with investigators and has placed him on leave.

[Newslit News @newslit]

Full story in today's Newslit Daily. https://t.co/0bhnWb1XkC

[Newslit News @newslit]

Apple is about to make Siri unrecognizable. iOS 27, previewed ahead of WWDC on June 8, will turn Siri into a standalone conversational app with persistent chat history — think ChatGPT, but built into every iPhone. The Camera app gets a dedicated AI mode powered by Visual Intelligence, and the Photos app gains three new generative editing tools.

Apple is using its next software cycle to close the AI gap with Google and OpenAI in one move.