The Archive

We're likely to see more price increases as the big AI companies plan to go public.

Notion's head of product said he was "astonished" at “the amount of people RT-ing this."

"Chat is dead" — at least, according to a senior OpenAI employee.

Aitana Lopez, AI avatar by creative agency The Clueless. | Image: The Clueless This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on AI confusion, follow Robert Hart. The Stepback arrives in our subscribers' inboxes at 8AM ET. Opt in for The Stepback here. How it started At first, AI influencers were relatively easy to identify - and to ignore. Aside from the occasional bursts of hype, they didn't seem to change much about the way social media worked. The earliest virtual influencers - Lil Miquela with her blunt fringe and freckles, Imma ...

How accurate does an AI system need to be?

Krishnan is reportedly starting a new institution to continue shaping Trump's AI policy.

President Donald Trump said he's discussing deals "where the American people can benefit from the success of AI."

A proposed $2 billion data center has become a political flashpoint in the small city of Shelbyville, Indiana. And the controversy has only grown more intense after the mayor, Scott Furgeson, was caught on camera saying of the "No Data Center" signs going up that, "I've seen a lot of these all over town, but I only see them in shitty houses," before adding, "most of them are rentals." The woman speaking to him in the clip quickly pushes back, saying that they're "working class," and someone chimes in to add something that a mayor shouldn't have to be told about their constituents: "it doesn't...

An AI-generated image of the royal family featuring two Queen Elizabeth IIs. | Image: Meta AI Facebook has long been filled with feeds of clickbait articles. Now, Meta is making its own clickbait articles with AI. The standalone Meta AI app now has a "For You" section that populates a list of clickbait-style stories for you to read. But the topics, images, and text are all AI-generated - and as questionable as you'd expect from AI-created works. The Meta AI app first launched in April 2025 with its focus on a public "Discover" feed that showed AI-generated images and conversations from other ...

Our first glimpse of the new AI Siri came all the way back at WWDC 2024. Apple has been on its back foot, AI-wise, for the past few years. But in a strange way, playing from behind might not be such a bad move. At WWDC on Monday, Apple appears to be getting ready to reintroduce us to the new Siri. Again. As a reminder, we met the new Siri in 2024 when Apple "launched" Apple Intelligence. Siri came with a new glowing border, different voice options, and the ability to punt questions to ChatGPT. The whole "Intelligence" bit of the Siri redesign was coming soon, Apple promised. It didn't. In fac...

Release: micropython-wasm 0.1a2 I added a CLI to micropython-wasm ( issue #7 ), inspired by the first draft of the blog entry when I realized it would be a great way to illustrate the Try it yourself section. Tags: python , sandboxing , webassembly , micropython

I've been experimenting with different approaches to running code in a sandbox for several years now, but my latest attempt feels like it might finally have all of the characteristics I've been looking for. I've released it as an alpha package called micropython-wasm , and I'm using it for a code execution sandbox plugin for Datasette Agent called datasette-agent-micropython . Why do I want a sandbox? My key open source projects - Datasette , LLM , even sqlite-utils - all support plugins. I absolutely love plugins as a mechanism for extending software. A carefully designed plugin system reduc...

OpenAI Help: Lockdown Mode OpenAI first teased this in February , but now it's live and "rolling out to eligible personal accounts, including Free, Go, Plus, and Pro, and self-serve ChatGPT Business accounts": Lockdown Mode is designed to help prevent the final stage of data exfiltration from a prompt injection attack by limiting outbound network requests that could transfer sensitive data to an attacker. Lockdown Mode does not prevent prompt injections from appearing in the content ChatGPT processes. For example, a prompt injection could appear in cached web content or in an uploaded file, a...

Applications for Startup Battlefield 200 officially close on June 8, 11:59 p.m. PT. Now’s not the time to wait any longer. Secure your shot at competing on the Disrupt Stage at TechCrunch Disrupt 2026 this October at San Francisco's Moscone West.

The companies announced the deal on Friday, just one week ahead of SpaceX's historic IPO.

Your broken harness is actively making the model worse. Here's what I keep seeing after years of eyeballing trajectories, and what you need to fix.

SpaceX won’t get easy access to billions of dollars from passive investors.

Developer felt "beaten up," with "no choice" but to shrink data center.

We investigate the probabilistic reasoning capabilities of large language models through a controlled benchmarking study on discrete probability problems. We constructed two datasets, respectively a set of standard exercises and a set of counterintuitive exercises, designed to trigger heuristic reasoning, and evaluated 8 state-of-the-art models, each tested with and without Chain-of-Thought prompting. Models achieve an average accuracy of 0.96 on standard problems but only 0.59 on counterintuitive ones. We further provide empirical evidence of token bias: performance drops by over 20% when ca...

Humans learn from social life. Simulating this process with LLM-powered agents represents a promising research direction, raising a natural question: whether LLMs can learn from such simulated social experience to better understand and replicate human behavior. However, prior agent society simulations typically operate at the scale of days, limiting the depth of social interactions and long-term growth. In this paper, we study long-term life simulation and LLM learning in agent societies, with two goals: (1) investigating social behaviors that emerge from life-long simulation, and (2) develop...

Current Vision-Language Models struggle with hours-long videos because processing full-length visual sequences induces prohibitive token explosion and attention dilution. To overcome this, we introduce MemDreamer to decouple perception and reasoning, shifting long-video understanding into an agentic exploration process. As a plug-and-play framework, it incrementally streams videos to construct a Hierarchical Graph Memory, a top-down three-tier architecture for semantic abstraction, anchored by a foundational graph capturing spatiotemporal and causal relations. During inference, the reasoning ...

Large language models exhibit impressive zero-shot capabilities across a wide range of downstream tasks. However, they struggle to function as off-the-shelf embedding models, leading to suboptimal performance on massive text embedding benchmarks. In this paper, we identify a potential cause underlying this deficiency. Our motivation stems from an unexpected observation: text embeddings tend to align with frequent but uninformative tokens when projected onto the vocabulary space. We argue that this excessive expression of high-frequency tokens suppresses the model's ability to capture nuanced ...