Hi there, this is your daily ☕️ Techpresso.
In today's Techpresso:
👷 Anthropic turns to ‘skills’ to make Claude more useful at work
👉 Apple readies its first ever touch-screen MacBook
🛑 OpenAI suspends Sora depictions of Martin Luther King Jr
🚓 Amazon Ring cameras are moving deeper into law enforcement
📉 AI bots and summaries hurt Wikipedia traffic
⚠️ Massive data leak of US officials
🎁 + 16 other news you might like
🔮 + 2 handpicked research papers and tools
👷 Anthropic turns to ‘skills’ to make Claude more useful at work LINK
Anthropic introduced Skills, which are folders containing a Markdown file and scripts that give Claude specific instructions for handling specialized tasks like creating documents or using certain tools.
To be token efficient, the model first reads a short explanation for each Skill from its frontmatter YAML, only loading the complete details when a user's request requires them.
Since the feature depends on having a filesystem to execute commands, anyone can build and distribute Skills as simple files, effectively turning Claude into a general automation agent.
👉 Apple readies its first ever touch-screen MacBook LINK
Apple is reportedly developing its first touchscreen MacBook Pro, which is expected to launch with the M6 chip sometime in late 2026 or early 2027 for a premium price.
This high-end laptop will reportedly mark the debut of an OLED screen, also swapping the controversial top notch for a more modern "hole-punch design" similar to the iPhone.
To address the common wobble issue on touch laptops, Apple is creating a "reinforced hinge and screen hardware" so the panel does not bounce back when you tap it.
🛑 OpenAI suspends Sora depictions of Martin Luther King Jr LINK
OpenAI suspended video generations of Martin Luther King Jr. in Sora at the request of his estate after people used the app to create disrespectful depictions.
The content that prompted the action included Sora-made videos of King making monkey noises during a speech and another showing him wrestling with civil rights leader Malcolm X.
The company is now strengthening its guardrails and allows representatives for other historical figures to ask for their likenesses not to be used in the AI videos.
🚓 Amazon Ring cameras are moving deeper into law enforcement LINK
Amazon's Ring is partnering with the AI camera network Flock, letting law enforcement agencies request doorbell video from users for evidence collection and their investigative work.
Flock's AI technology scans license plates and lets government customers search footage for people by description; agencies like ICE and the Secret Service already have access.
The deal could expand Flock's reach to millions more cameras, despite Ring's past FTC order over employees having unrestricted access to customers' private videos for years.
📉 AI bots and summaries hurt Wikipedia traffic LINK
After making its bot detection more accurate, Wikimedia's data reveals an eight percent year-over-year drop in page views, a decline it attributes to generative AI search summaries.
The foundation warns this traffic decline could lead to fewer volunteers, less funding for the site, and ultimately a drop in the quality and reliability of its content.
Wikimedia is asking LLMs and search engines to clearly state where information is sourced from and make it easier for people to visit and contribute to the original articles.
⚠️ Massive data leak of US officials LINK
A hacking group called “Scattered LAPSUS$ Hunters” publicly released names, phone numbers, and residential addresses for hundreds of federal government employees at the DHS, FBI, and DOJ.
The dump contained spreadsheets detailing the information of 680 DHS officials and over 170 FBI email addresses, with some data corroborated by a cybersecurity firm.
The hackers posted taunting messages referencing Mexican cartels and money, while also threatening to release sensitive data in the future belonging to the Internal Revenue Service.
Other news you might like
- More than half of new content is AI-generated now, report findsLINK
- China Hackers Test AI-Optimized Attack Chains in TaiwanLINK
- Pinterest adds controls to let you limit the amount of ‘AI slop’ in your feedLINK
- Uber will offer gig work like AI data labeling to drivers while not on the roadLINK
- Waymo dips its wheels back into delivery, this time with DoorDashLINK
- Anthropic connects Claude to Microsoft Teams, Outlook, and OneDriveLINK
- Snap is bringing shopping features to its AR glassesLINK
- Coinbase Launches Stablecoin Payments Platform for BusinessesLINK
- New York bans AI-enabled rent price fixingLINK
- Big Tech is paying millions to train teachers on AILINK
- Reddit's AI Suggests Users Try HeroinLINK
- Hollywood Agents Seethe Over Sora 2, Say OpenAI Purposely Misled ThemLINK
- Google Quietly Starts Rolling Out Gemini 3.0 Pro Ahead of Official LaunchLINK
- Kayak launches an ‘AI Mode’ for travel questions, search, and bookingsLINK
- Amazon shares a ‘first look’ at new nuclear facilityLINK
- Helion lands permit for constructing reactor building at Washington site that aims to be first fusion plantLINK
Latest research and tools
Datapizza AI: an open-source framework for building and debugging generative AI applications that use multiple agents or retrieve information from documents.LINK
Every Language Model Has a Forgery-Resistant Signature: this paper shows that all AI language models leave a unique and difficult-to-forge statistical signature on the text they produce.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!