While there are a lot of copyright circumstances working their manner by means of the courtroom system, we now have an essential resolution from considered one of them. Judge William Alsup dominated that the usage of copyrighted materials for coaching is “transformative” and, therefore, honest use; that changing books from print to digital type was honest use; however that the usage of pirated books in constructing a library for coaching AI was not.
Now that everybody is attempting to construct clever brokers, we’ve to suppose severely about agent safety—which is doubly problematic as a result of we already haven’t thought sufficient about AI safety and points like immediate injection. Simon Willison has coined the time period “lethal trifecta” to explain the mix of issues that make agent safety notably tough: entry to non-public knowledge, publicity to untrusted content material, and the flexibility to speak with exterior providers.
Artificial Intelligence
- Researchers have fine-tuned a mannequin for finding deeds that embrace language to forestall gross sales to Black individuals and different minorities. Their analysis exhibits that, as of 1950, roughly 1 / 4 of the deeds in Santa Clara county included such language. The analysis required analyzing thousands and thousands of deeds, many greater than might have been analyzed by people.
- Google has launched its stay music mannequin, Magenta RT. The mannequin is meant to synthesize music in actual time. While there are some restrictions, the weights and the code can be found on Hugging Face and GitHub.
- OpenAI has discovered that fashions that develop a misaligned persona will be retrained to carry their habits again inline.
- The Flash and Pro variations of Gemini 2.5 have reached normal availability. Google has additionally launched a preview of Gemini 2.5 Flash-Lite, which has been designed for low latency and price.
- The web site lowbackgroundsteel.ai is meant as a repository for pre-AI content material—i.e., content material that might not have been generated by AI.
- Are the drawbridges going up? Drew Breunig compares the present state of AI to Web 2.0, when firms like Twitter began to limit builders connecting to their platforms. Drew factors to Anthropic reducing off Windsurf, Slack blocking others from looking out or storing messages, and Google reducing ties with Scale after Meta’s funding.
- Simon Willison has coined the phrase “lethal trifecta” to explain harmful vulnerabilities in AI brokers. The deadly trifecta arises from the mix of personal knowledge, untrusted content material, and exterior communication.
- Two new papers, “Design Patterns for Securing LLM Agents Against Prompt Injections” and “Google’s Approach for Secure AI Agents,” tackle the issue of immediate injection and different vulnerabilities in brokers. Simon Willison’s summaries are wonderful. Prompt injection stays an unsolved (and maybe unsolvable) drawback, however these papers present some progress.
- Google’s NotebookLM can flip your search outcomes right into a podcast based mostly on the AI overview. The characteristic isn’t enabled by default; it’s an experiment in search labs. Be cautious—listening to the outcomes could also be enjoyable, but it surely takes you farther from the precise outcomes.
- AI-enabled Barbie™? This I’ve to see. Or perhaps not.
- Institutional Books is a 242B token dataset for coaching LLMs. It was created from public area/out-of-copyright books in Harvard’s library. It contains over 1M books in over 250 languages.
- Mistral has launched their first reasoning mannequin, Magistral, in two variations: a Small model (open supply, 24B) and a closed Medium model for enterprises. The announcement stresses traceable reasoning (for purposes like legislation, finance, and healthcare) and creativity.
- OpenAI has launched o3-pro, its latest high-end reasoning mannequin. (It’s in all probability the identical mannequin as o3, however with totally different parameters controlling the time it might probably spend reasoning.) LatentSpace has a very good put up on the way it’s totally different. Bring a number of context.
- At WWDC, Apple introduced a public API for its on-device basis fashions. Otherwise, Apple’s AI-related bulletins at WWDC are unimpressive.
- Simon Willison’s “The Last Six Months in LLMs” is value studying; his private benchmark (asking an LLM to generate a drawing of a pelican driving a bicycle) is surprisingly helpful!
- Here’s an outline of device poisoning assaults (TPA) towards methods utilizing MCP. TPAs have been first described in a put up from Invariant Labs. Malicious instructions will be included within the device metadata that’s despatched to the mannequin—often (however not solely) within the description area.
- As a part of the New York Times copyright trial, OpenAI has been ordered to retain ChatGPT logs indefinitely. The order has been appealed.
- Sandia’s new “brain-inspired” supercomputer, designed by SpiNNcloud, is value watching. There’s no centralized reminiscence; reminiscence is distributed amongst processors (175K cores in Sandia’s 24-board system), that are designed to imitate neurons.
- Google has up to date Gemini 2.5 Pro. While we wouldn’t usually get that enthusiastic about an replace, this replace is arguably one of the best mannequin out there for code technology. And an much more spectacular mannequin, Gemini Kingfall, was (briefly) seen within the wild.
- Here’s an MCP connector for people! The thought is straightforward: When you’re utilizing LLMs to program, the mannequin will usually go off on a tangent if it’s confused about what it must do. This connector tells the mannequin easy methods to ask the programmer each time it’s confused, protecting the human within the loop.
- Agents look like much more weak to safety vulnerabilities than the fashions themselves. Several of the assaults mentioned on this paper contain getting an agent to learn malicious pages that corrupt the agent’s output.
- OpenAI has introduced the provision of ChatGPT’s Record mode, which data a gathering after which generates a abstract and notes. Record mode is presently out there for Enterprise, Edu, Team, and Pro customers.
- OpenAI has made its Codex agentic coding device out there to ChatGPT Plus customers. The firm’s additionally enabled web entry for Codex. Internet entry is off by default for safety causes.
- Vision language fashions (VLMs) see what they need to see; they are often very correct when answering questions on photographs containing acquainted objects however are very prone to make errors when proven counterfactual photographs (for instance, a canine with 5 legs).
- Yoshua Bengio has introduced the formation of LawZero, a nonprofit AI analysis group that can create “safe-by-design” AI. LawZero is especially involved that the most recent fashions are displaying indicators of “self-preservation and deceptive behavior,” little question referring to Anthropic’s alignment analysis.
- Chat interfaces have been central to AI since ELIZA. But chat embeds the outcomes you need, in a number of verbiage, and it’s not clear that chat is in any respect applicable for brokers, when the AI is kicking off a number of new processes. What’s past chat?
- Slop forensics makes use of LLM “slop” to determine mannequin ancestry, utilizing methods from bioinformatics. One result’s that DeepSeek’s newest mannequin seems to be utilizing Gemini to generate artificial knowledge fairly than OpenAI. Tools for slop forensics can be found on GitHub.
- Osmosis-Structure-0.6b is a small mannequin that’s specialised for one process: extracting construction from unstructured textual content paperwork. It’s out there from Ollama and Hugging Face.
- Mistral has introduced an Agents API for its fashions. The Agents API contains built-in connectors for code execution, internet search, picture technology, and various MCP instruments.
- There is now a database of courtroom circumstances by which AI-generated hallucinations (citations of nonexistent case legislation) have been used.
Programming
- Martin Fowler and others describe the “expert generalist” in an try to counter rising specialization in software program engineering. Expert generalists mix one (or extra) areas of deep data with the flexibility so as to add new areas of depth shortly.
- Duncan Davidson factors out that, with AI in a position to crank out dozens of demos in little time, the “art of saying no” is immediately essential to software program builders. It’s too straightforward to get misplaced in a flood of respectable choices whereas attempting to choose one of the best one.
- You’ll in all probability by no means have to compute a billion factorials. But even for those who don’t, this text properly demonstrates optimizing a tough numeric drawback.
- Rust is seeing elevated adoption for knowledge engineering initiatives due to its mixture of reminiscence security and excessive efficiency.
- The finest method to make programmers extra productive is to make their job extra enjoyable by encouraging experimentation and relaxation breaks and taking note of points like applicable tooling and code high quality.
- What’s the subsequent step after platform engineering? Is it platform democracy? Or Google Cloud’s new thought, inside improvement platforms?
- A research by the Enterprise Strategy Group and commissioned by Google claims that software program builders waste 65% of their time on issues which might be solved by platform engineering.
- Stack Overflow is taking steps to protect its relevance within the age of AI. It’s contemplating incorporating chat, paying individuals to be helpers, and including personalised residence pages the place you possibly can mixture essential technical info.
Web
- Is it time to implement HTTP/3? This customary, which has been round since 2022, solves a number of the issues with HTTP/2. It claims to scale back wait and cargo instances, particularly when the community itself is lossy. The Nginx server, together with the main browsers, all help HTTP/3.
- Monkeon’s WikiRadio is an internet site that feeds you random clips of Wikipedia audio. Check it out for extra initiatives that remind you of the times when the net was enjoyable.
Security
- Cloudflare has blocked a DDOS assault that peaked at 7.3 terabits/second; the height lasted for about 45 seconds. This is the most important assault on report. It’s not the type of report we prefer to see.
- How many individuals do you guess would fall sufferer to scammers providing to ghostwrite their novels and get them printed? More than you’d suppose.
- ChainLink Phishing is a brand new variation on the age-old phish. In ChainLink Phishing, the sufferer is led by means of paperwork on trusted websites, well-known verification methods like CAPTCHA, and different reliable sources earlier than they’re requested to surrender non-public and confidential info.
- Cloudflare’s Project Galileo provides free safety towards cyberattacks for weak organizations, similar to human rights and aid organizations which might be weak to denial-of-service (DOS) assaults.
- Apple is including the flexibility to switch passkeys to its working methods. The capability to import and export passkeys is a crucial step towards making passkeys extra usable.
- Matthew Green has a wonderful put up on cryptographic safety in Twitter’s (oops, X’s) new messaging system. It’s value studying for anybody fascinated about safe messaging. The TL;DR is that it’s higher than anticipated however in all probability not so good as hoped.
- Toxic agent flows are a brand new type of vulnerability by which an attacker takes benefit of an MCP server to hijack a consumer’s agent. One of the primary situations pressured GitHub’s MCP server to disclose knowledge from non-public repositories.
Operations
- Databricks introduced Lakeflow Designer, a visually oriented drag-and-drop no code device for constructing knowledge pipelines. Other bulletins embrace Lakebase, a managed Postgres database. We have at all times been followers of Postgres; this can be its time to shine.
- Simple directions for making a bootable USB drive for Linux—how quickly we overlook!
- An LLM with a easy agent can vastly simplify the evaluation and analysis of telemetry knowledge. This shall be revolutionary for observability—not a risk however a possibility to do extra. “The only thing that really matters is fast, tight feedback loops.”
- DuckLake combines a standard knowledge lake with an information catalog saved in an SQL database. Postgres, SQLite, MySQL, DuckDB, and others can be utilized because the database.
Quantum Computing
- IBM has dedicated to constructing a quantum laptop with error correction by 2028. The laptop could have 200 logical qubits. This in all probability isn’t sufficient to run any helpful quantum algorithm, but it surely nonetheless represents an enormous step ahead.
- Researchers have claimed that 2,048-bit RSA encryption keys might be damaged by a quantum laptop with as few as 1,000,000 qubits—an element of 20 lower than earlier estimates. Time to implement postquantum cryptography!
Robotics
- Denmark is testing a fleet of robotic sailboats (sailboat drones). They’re meant for surveillance within the North Sea.