Feed aggregator

To demonstrate the capabilities of its next-generation Muse Spark model, Meta has released a terminal coding agent called Muse Code that it thinks can help developers to tidy up their software projects. Meta co-trained Muse Code on version 1.2 of its Muse Spark model, also released this week and now apparently boasting improved code generation smarts. Developers can think of this beta release as the equivalent to OpenAI Codex or Anthropic’s Claude Code, two other LLM-based service offerings tweaked for the modern coder. Meta designed its new agent to be handy at planning changes to a codebase, writing the code and validating the results. Currently, Meta has Muse Spark locked away as a proprietary, closed-weight model hosted in the cloud, an approach its rivals also embrace but which departs from the open-weight approach Meta previously implemented with its Llama models. But Meta CEO Mark Zuckerberg did not rule out opening up Muse Spark in the future. “I'll have more to share on that soon,” he replied to a question posed on X about Muse Spark being open source. An agent orchestrator Muse Code is best described as an agent orchestrator that runs on your command line. As Zuck noted in a series of X messages, when a developer starts a task, Muse Code fires up background agents to maintain a context file that other sub-agents doing the work can consult should they lose their way. The tool logs every action before execution, so no work is lost. Multiple agents can work in parallel on the task using their own isolated work trees. “Your working copy is never touched,” Zuck wrote. In one test, the agent platform simultaneously built six features for a single game, with no collisions among the agents, Zuckerberg enthused. “TBH it's a good harness,” boasted Hongyu Ren, a researcher for Meta’s Superintelligence Labs, on X. Remedial coding Muse Code relies on Meta’s Muse Spark Large Language Model (not to be confused with the Apache Spark big data cruncher). Muse Spark 1.2 is the third release in four months from Meta Superintelligence Labs, a unit that Meta stood up in June 2025 to reinvigorate the company’s AI efforts and pursue creation of a personalized AI “superintelligence” that focuses on deep reasoning and long-horizon planning. The first model from this group, Muse Spark, is a multi-modal model able to digest and reason against text, images, video, audio, and even PDFs. It supports agents in long-running tasks. With the first release of the model in April, Meta boffins admitted in the announcement that they needed to work more on Spark’s coding abilities. The new 1.2 release addressed that deficiency. In another test, Muse Spark, running on Nvidia Hopper GPUs, tackled a kernel optimization task. Its agents made over 1,000 tool calls over a 24-hour period. “It kept finding substantial improvements well beyond the initial exploration phase,” Zuckerberg wrote. Understand the mission In his X missives, Zuckerberg included a somewhat vague chart comparing the performance of Muse Spark against other commercial models, using two industry benchmarks – Terminal-Bench 2.1 and DeepSWE 1.1 – that evaluate how autonomous agents act like software engineers, as well as a Meta Internal Coding bench too. The benchmarks all showed Muse Spark to be close to the best, but never the very best, at understanding the assigned engineering task and executing it with as little mission creep as possible. Muse Spark performed honorably compared to Opus 5, GPT5.6 Terra, Grok 4.5 and Gemini 3.6. The scores are tightly clustered, so they all did well (except occasionally Gemini, the current laggard du jour). So, Muse Spark is competitive anyway, though one eagle-eyed commenter wondered why OpenAI’s midline GPT5.6 Terra was used, instead of the top-tier GPT5.6 Sol. Installing the agent within your command line is possible through a curl command. Poly-model enthusiasts can also tap into Muse Spark via OpenRouter, or through its API. Muse Spark’s actual intelligence is metered at US$1.25 per million input tokens and US$4.25 per million output tokens. Discounts are available and the service offers a respectable 1 million token context window. ®

If you want to understand just how racist Republican voters are, look no further than Tuesday night’s result in Michigan’s 8th Congressional District primary. Both President Donald Trump and the National Republican Congressional Committee had endorsed veteran Amir Hassan to face off against Democratic Rep. Kristen McDonald Rivet in this seat that Trump carried by 2 points in 2024.

Source

A cartoon by Pedro Molina. Related | Looks like Trump will get Todd Blanche after all…

Source

Santa Cruz County released its 2026 point-in-time count results on Wednesday, which show an increase in homelessness over 2025, but a level still well beneath the 2022 peak.

It’s painfully evident that Republicans have a very cramped idea of what religious freedom is, one that protects only their very narrow, very weird, very deadly, very stupid brand of evangelical Christianity. The latest example of this is Florida Attorney General James Uthmeier, who demanded Catholic schools to comply with state law to grant vaccine exemptions or lose access to funding.

Source

Dominican Hospital earned a top spot among 411 California hospitals after it was rated high performing in eight adult procedures and conditions. U.S. News & World Report named it a “Best Regional Hospital.”

Editor’s note: This is part 4 of Tarmo Hannula’s recent trip to Los Angeles

Packing up and heading south on day five of our road trip to southern California, my wife Sarah and I rolled through a number of towns and cities on our way to Carlsbad.

On our way out of Los Angeles we stopped for lunch at Little Saigon in Orange County about 45 miles south of downtown LA. It’s the largest Vietnamese enclave outside of Vietnam. One shopping center after another featured stores and restaurants with tons of Vietnamese goods and food.

COASTLINE Highway 1 stretches along the coast near Malibu.

We made a brief stop in south LA to view an outdoor sculpture park called Sister Dreamer that Sarah had read about. The massive sculptures and towering pillars echo Egyptian monuments and were created by artist Lauren Halsey who was born in the area. Halsey said the monuments are dedicated to the people in the neighborhood to “honor local heroes, neighbors and family.” 

LOCALS HONORED Large sculptures are shown in southern Los Angeles at the Sister Dreamer sculpture park.

In Huntington Beach we discovered a great birding spot, the Bocha Chica National Reserve along Highway 1. The kind woman at the Nature Center, Jess, told us about a spot where a group of least terns were nesting and laying eggs. On our way there we spotted a giant sea turtle treading the shallow waters of a lagoon. Then we saw a yellow-crowned night heron, a first for both of us. Black skimmers and great white herons also dotted the water’s edge. There were about a dozen least terns sitting on their nests on the shoreline in a protected area.

IN THE WILD A yellow-crowned night heron works the sands of the Bocha Chica National Reserve. 

As with many of our road trips, we next aimed for the 1776 California Mission at San Juan Capistrano. According to their website it is “known as the ‘Jewel of the Missions.’” The site features a museum with permanent exhibits with original artifacts as well as traveling and temporary exhibits. The site was founded on Nov. 1, 1776, by Junípero Serra as the seventh of 21 California missions established by Spain.

In the next part of this series we settle in at the Great Western Hotel right on the rocky shore of Carlsbad, our furthest point south, to visit family and continue exploring.

OPINION In a post-apocalyptic setting – a zombie outbreak, for example – a small, off-grid language model running on something like a Raspberry Pi sounds like a survival cheat code. It’s certainly a clever way to get folks to check out your physical AI platform and maybe even splash out 7x the MSRP for an Nvidia Jetson Orin Nano with a battery enclosure, as one startup has done. But the scenario also sets a dangerous precedent, and in a real-world survival situation that local LLM could end up being a bigger liability than a lifesaver. Roleplaying a zombie apocalypse with a local AI model is harmless fun. Zombies aren’t real and you’re probably not LARPing your way through the backcountry pursued by the undead. The problem is there’s a certain group of people who are going to see the concept presented in videos like this and think an off-grid AI is a great idea for a bug-out bag or to bring along on a hiking trip, and it’s not hard to see why. Bit-for-bit, these models’ ability to retain information is unrivaled. You could download the entire English text of Wikipedia (about 24.7 GB compressed), and still not approach the information density of a four-billion-parameter model taking up 2 GB of memory. Add in vision capabilities, support for translation, and maybe layer in a few text-to-speech and speech-to-text models and it starts to sound like the perfect survival companion. But for reasons you have probably already experienced over the past four years, we wouldn’t recommend it. Even the largest LLMs still hallucinate — that is, they’ll confidently fabricate information and present it as fact. That’s because while models are trained on tens of trillions of words (really, tokens) of data, their ability to retrieve that data is inherently imperfect. LLMs don’t actually know what words belong together, just that statistically one word or another has the highest probability of coming next. It’s like auto-complete on steroids. Modern chatbots from OpenAI, Anthropic, or Google attempt to mitigate potential hallucinations by pulling in live web data to better inform their responses. The model is no longer answering your query on its own, but rather conducting research and presenting its findings. Even then, there’s no guarantee the information it retrieved is accurate or applicable to your specific context. Medical professionals have been sounding the alarm on the dangers of relying on LLMs for medical advice for some time now. According to OpenAI, more than 230 million people globally query ChatGPT with health and wellness related questions each week. The problem, medical professionals at Duke University School of Medicine note, is that these models can’t read between the lines to understand what the patient is really trying to say, and often suffer from sycophantic tendencies that lead them to tell users what they want to hear. “When a patient comes to us with a question, we read between the lines to understand what they’re really asking,” Ayman Ali, a fourth-year surgical resident at Duke Health, said in a blog post earlier this year. “We’re trained to interrogate the broader context. Large language models just don’t redirect people that way.” Off grid, an LLM can’t fall back on the web for help and almost certainly doesn’t benefit from the same guardrails that major chatbots use to protect users against potentially dangerous outputs. As we’ve shown time and time again, getting local LLMs running on inexpensive hardware has never been easier. For a contrived scenario like roleplaying a zombie apocalypse, the risk is minimal. But change things up and ask the model whether the berries pictured are edible, and you could be playing Russian roulette with a probability engine. ®

A Santa Cruz judge postponed seafood entrepreneur Clarice Owens’ felony stalking trial for a third time, pushing the start date from Aug. 17 to Nov. 30 after her attorney requested a delay.

When it comes to movies, everyone knows the sequel is never as good as the original. But what happens when nobody liked the original to begin with? Normally this would prompt producers to put a halt to any future installments. But first lady Melania Trump’s self-titled documentary “Melania” is an apparent exception. Yes, Jeff Bezos’ Amazon MGM Studios has decreed the ’s film flop means…

Source

ph @ross_inside

One of Elon Musk’s wayward rockets seems to have crashed into the moon, but the Trump administration isn’t worried about it.

Part of a SpaceX rocket crashed into the moon Wednesday morning, reportedly with the impact of three tons of TNT. The crash sent up a big cloud of dust and left a new crater on the moon’s surface, The Independent reports. Images of the impact may take days to appear from ground telescopes and satellites orbiting the moon. The SpaceX rocket is from a Falcon 9 spacecraft that was carrying a lunar lander to the Moon last January.

When asked about the explosion on Fox News, NASA Administrator Jared Isaacman was dismissive, crediting SpaceX with recovering 600 of its rockets in other cases, and said this explosion wasn’t a big deal in terms of space trash.

“It’s very infrequent to have things like the Falcon 9 second stage crash into the moon,” Isaacman said. “It’s not a big deal right now, the moon has clearly seen better days.”

NASA Administrator Jared Isaacman on rockets crashing into the moon: "It's not a big deal right now. The moon has clearly seen better days." pic.twitter.com/IiNt7PdM2V

— Aaron Rupar (@atrupar) August 5, 2026

Isaacman then pivoted into how once reusable rockets are ready, they would be landing on the moon to help “contribute to the city block you’re trying to build on the surface.”

Isaacman has a close business relationship with Musk, having financed SpaceX missions in the past, and wants to carry out Musk and President Trump’s dreams of building a base on the moon. If there were major problems with spacecraft crashing into the moon, he’s not likely to hurt his own position, or make Musk look bad, by telling the public in a news interview.

Space debris is a growing problem from increased rocket launches, and could become a major issue if any important satellites are damaged. If more of Musk’s SpaceX rockets veer out of control, his moon base plans could hit a major snag.

A candidate who suspended his campaign a month ago, spent no money, and endorsed his own rival won Michigan’s 8th Congressional District GOP primary on Tuesday, handily defeating Amir Hassan, the candidate President Donald Trump endorsed.

The victor, Thomas J. Smith, won 50 percent of the vote in the state’s 8th congressional district, according to the New York Times, while Amir Hassan only had about 33 percent. That’s a difference of over 10,000 votes. Smith suspended his campaign in mid-July—too late to get his name off the ballot. In a July 16 video, he formally endorsed Al Lemmo, donning an “Al Lemmo for Congress” hat. Lemmo ended up with 16 percent of the vote on Tuesday.

Trump endorsed Hassan on Truth Social in June, but supporters opposed the move, some with significant anti-Islamic sentiment. Hassan’s campaign website says the congressional candidate “recently reaffirmed his commitment to Christ” and stated that “Jesus Christ is his Lord and Savior.” Hassan also received support from House Republicans’ campaign committee, which highlighted his service in the Navy and as a former federal law enforcement officer. 

According to FEC campaign filings, Hassan spent over $600,000, while Smith has no FEC filings. 

According to the Detroit News, Smith spoke for about 13 seconds at a Wednesday Republican event at a restaurant in Warren, Michigan: “I’m running for the 8th Congressional District to unseat communist comrade Rivet.”

In a statement to Politico earlier on Wednesday, Smith said: “I look forward to working with Speaker [Mike] Johnson and the entire Republican team to expand our House majority and advance President Trump’s America First agenda to lower costs for working families and make America affordable again.”

Smith is set to face Kristen McDonald Rivet, the Democratic incumbent who has served as the representative for the district since 2025.

Nearly a dozen flaws, some critical, in major AI agent frameworks that enterprises use to build apps reveal a security failure that extends beyond prompt injection - or any single model - according to Check Point researchers. “Our research shows a deeper failure: in many agentic frameworks, prompt-controlled content can cross the boundary into trusted framework logic itself,” Yarden Porat and Shahar Tal note in a write-up about a Wednesday Black Hat talk on post-injection exploitation across AI agent frameworks, which they also discussed with The Register. “A bug in an agent framework isn't a bug in one product - it's a bug in the layer a whole category of AI apps runs on,” Tal told us. “And the agent needs no dangerous tools to be turned against you: reading the wrong document is enough. We’re building this layer faster than we know how to defend it.”
 The researchers spent a year trying to break various frameworks that enterprises use including LangChain, LangGraph, CrewAI, AutoGen, Microsoft Agent Framework, and Google ADK. And across these frameworks, the team found and disclosed 11 vulnerabilities. “Almost none of it was a completely new bug class,” Tal said. “That's insecure deserialization, server-side request forgeries, path traversals, use-after-free. These are bugs that we learned to fix 20 years ago, and they're sitting underneath agents that now read your inbox, or update your database.” These are old types of threats, and the model isn’t the weak link, he added. The failure exists in the “plumbing around the model, and we think this has been overlooked,” Tal told us. “There’s a lot of research going into prompt injection and defenses, which are important, but that’s just the beginning.” Defenders should assume prompt injection, according to the researchers. The bug is what the framework does with the injection - and in these cases, the threat hunters found that the frameworks often fail to keep attacker-controlled content in the data plane. This allows it to influence trusted orchestration, memory, state, routing, and system instructions. For example, the duo found a critical checkpoint deserialization bug in Microsoft Agent Framework that led to remote code execution. “Agents have checkpoints, which are a way for them to save their state or rewind to an earlier point,” Tal explained. These checkpoints are saved snapshots of an agent's state, or task progress at a specific moment, and they serialize data - such as conversation history - into persistent storage, so if an error occurs, the system reloads this saved state instead of starting from scratch. In this case, Check Point’s team found an insecure deserialization issue where, via prompt injection, the agent loaded untrusted checkpoint data, and this could allow attackers to execute malicious code on the system. “One person's message plants the payload, and then a different person rewinds their own session, which triggers the payload, and now the attacker has a shell on that server,” Tal said. Microsoft recognized the researchers’ findings, paid a $10,000 bug bounty and fixed the issue. But because the framework wasn’t a generally available product when Check Point found the flaw, Microsoft did not issue a CVE. Microsoft told us that it appreciated the researchers reporting the vulnerability. “We have released protections to harden the Agent Framework and prevent the concrete exploitation path demonstrated in the proof of concept,” a spokesperson told The Register. “In addition, we updated the specific checkpoint file with additional language to define the security boundary.” The duo also found flaws in Google ADK (agent development kit). However, Google responded differently, the researchers told us, and did not completely fix the vulnerability or issue a CVE. “ADK ships a built-in development assistant that can write files, and it stays reachable over the HTTP API even though it is hidden from the app listing,” Porat told us. To break this trust boundary, an attacker opens a session, asks ADK to write an agent whose Python code runs at import time, and then asks the server to run the agent, he explained. The server then imports the file and executes the attacker’s code. “There is no authentication on that API by default, and adk deploy cloud_run publishes the same API, so on a default Cloud Run deployment it is reachable without credentials,” Porat said. “From there it reaches the environment's API keys and the container's Google Cloud service account." Google did not respond to The Register’s inquiries. But according to Check Point, Google initially deemed the issue not a bug. “We argued the consequence rather than the mechanism: code execution on that container reaches the environment's API keys and the container's Google Cloud service account, which is secret theft, not a developer inconvenience,” Porat said. Google ultimately paid a $3,133.70 bounty and issued a partial fix, we’re told. In total, the bug hunters received $17,133.70 in rewards for their efforts. And this isn’t a story about one vendor or framework doing a “particularly bad job,” Tal said. “If one was an outlier, this would be a story about that one vendor,” he added. “Our finding is that the same bug classes turn up in all of them.” ®

The serial entrepreneur is stepping down a little over a year after taking the "24/7 job" of overseeing X.

Kalanick continues to get the band back together, after acquiring Anthony Levandowski's autonomy startup, and even soliciting investment from Uber itself.

Meta expanded its AI coding offerings with a new agent that, it promises, can handle complex tasks with complex software.

The DOJ alleged that OpenAI did not meaningful attempt to hire U.S. citizens before seeking permanent residence for Visa-holding employees.

⚡ Quick Take

• In Spider-Man: Brand New Day, Peter Parker encountered several iconic Marvel heroes. But who should Spidey share the big screen with next time?

Spider-Man: Brand New Day is breaking box office records left and right, and the world is in the grip of Spidey-Mania. This almost guarantees another Spidey installment starring Tom Holland and set in the MCU. And like the previous Marvel Studios-produced Spider-Man films, contractually Sony must include one Marvel Studios hero, quite possibly two. Given those rules, here are the Marvel Comics heroes we really hope share the screen with Peter Parker in future installments.

Johnny Storm/The Human Torch Sony/Marvel Studios/Marvel Comics

One of the longest-running bromances in Marvel Comics is nerdy dork Peter Parker and hotshot ladies’ man Johnny Storm, a.k.a. the Human Torch. These two have been giving each other crap (in a loving way) for six decades, and despite all the insults they hurl at each other, they are truly the best of buds. Currently in the MCU, Johnny (Joseph Quinn) lives in another reality, Earth-828. But we have a feeling once Avengers: Secret Wars is over, they’ll be living on the same Earth. Of course, we’d like to see the full Fantastic Four meet Spidey, just as they did in Amazing Spider-Man #1. But for us, it’s really all about Peter and Johnny sitting on rooftop, discussing their rotten romantic luck together.

Matt Murdock/Daredevil Marvel Studios/Marvel Comics/Sony Pictures

Yeah, yeah, we know. Matt Murdock already appeared in cameo form in No Way Home. But that Charlie Cox appearance was just a tease. We’re dying to see Marvel’s two biggest street-level heroes have a real adventure together on the big screen. We were so sure that Brand New Day was hiding a Daredevil cameo. Simply because we just had a hard time imagining Spidey fighting the Hand without ol’ horn head. But he did. So we still need to see these two take on Kingpin, who began his comics existence as a Spidey villain. Tom Holland can’t finish out his MCU trilogy without fighting alongside the Man Without Fear. At least, we certainly hope not.

Jessica Drew/Spider-Woman Marvel Comics

Here’s a major Marvel hero that Sony and Marvel share the rights to, but shockingly has yet to appear in live-action. We’re talking about Jessica Drew, the original Spider-Woman. Unlike later heroes with that name, Jessica’s arachnid powers are not an exact mimic of Peter Parker’s, and her origins are wildly different. She’s a former SHIELD agent, ally of Nick Fury, and an Avenger. We know a Spider-Woman movie was in development for years, but in the wake of both Madame Web and Supergirl flopping, we doubt the studios would be brave enough to give Jess her own film in this climate. So please, find space for her in one of the two upcoming Spider-Man sequels. It’s long past time she got her due.

Bobby Drake/Iceman Twentieth Century Studios/Marvel/Sony Pictures

The only iteration of the mutant hero Iceman we’ve had on screen is Shawn Ashmore in the Fox X-Men films. And so far, there are no hints that Bobby Drake will appear in the upcoming X-Men reboot. So why not introduce him in a Spidey movie? Peter Parker has a long friendship with Bobby in other media, mostly in animated form, like the ‘80s cartoon Spider-Man and His Amazing Friends. In a Spider-Man movie, he can be a new “amazing friend” for Peter, who is finally reconnecting with other people again. Also, it would be great to finally have another young out gay hero in the MCU, besides Wiccan. It’s long overdue.

Kamala Khan/Ms. Marvel Sony Pictures/Marvel Comics/Marvel Studios

When Ms. Marvel was created back in 2014, the intent was for her to be the Spider-Man for a new generation. In essence, a teen hero juggling super-heroism with family and romantic drama. And both the comic book and MCU version, played by Iman Vellani, fully succeeded at that archetype. Sadly, we haven’t seen Iman Vellani as Jersey City’s Kamala Khan since The Marvels. And since there’s likely not ever going to be a follow-up to that film, we’d love to see her show up in a Spidey sequel. Peter Parker would be a great mentor figure, especially since he’s one of the few people who have lived through what she’s been through as a teenager. Time for Peter to take on the fatherly/big brother role for a change.

Luke Cage Sony Pictures/Marvel Comics/Marvel Television

Much like Daredevil, Luke Cage is another iconic Marvel street-level hero who has a long-time friendship with Spider-Man in the comics. Both heroes became close when they served together on the New Avengers, but go back even before that. Cage’s wife, Jessica Jones, even went to high school with Peter Parker. However, for the MCU, that timeline doesn’t add up. Mike Colter is already returning as the bulletproof hero of Harlem in the next season of Daredevil: Born Again. But after that, we hope he appears in a future Spider-Man installment. We feel Luke would get kind of irritated by Peter, but in a way that he’d also find endearing.

There are plenty of other heroes from the MCU we’d love for Spider-Man to team up with, like Ant-Man, and even Deadpool. But these we mentioned we see as essential team-ups. In the meantime, you can watch Spidey together with the Punisher, the Hulk, and Black Widow in Spider-Man: Brand New Day, now in theaters.

The post MCU Heroes We Hope Share the Screen with Spider-Man appeared first on Nerdist.

AI webpage crawlers are now being served their very own ads that ordinary visitors never see, with one firm's boss openly describing a strategy to influence what chatbots say about brands. The next time you ask Claude about where to bank, its answer may have been influenced by this bot-targeted content. We only have one documented example so far, and that's Time serving AI-only ads to selected AI crawlers, as spotted by Germany-based freelance software developer Vincent Schmalbach. In a Wednesday blog post, Schmalbach detailed how he found sponsored content embedded in markdown versions of some Time pages that are served to AI crawlers but not ordinary browsers. Those pages also contained advertising tags from adtech vendor Mobian ahead of extensive FAQs for online-only bank Ally. The FAQs include brand "facts," such as the number of fee-free ATMs associated with the branchless bank, alongside claims that Ally is "the only bank built for life today," putting it in "a category of one." Key “brand fact” statements that make for great regurgitable facts are also included in the AI-only advertisement, as are answers to questions like “is Ally good for everyday banking,” “can you deposit cash at Ally Bank,” and the like. The FAQ is flagged as sponsored content on the markdown page, but it doesn’t change the fact that the entire thing is written like it was designed to be spit back out by a chatbot in response to specific questions. Not all web crawlers are getting these markdown ads either. Per Schmalbach, Google’s standard search indexing bot doesn't see the ads, as it just gets the same HTML that a human gets. “Only the [chatbot] assistant crawlers get the forked version,” he wrote. That means user-agents including ClaudeBot, OAI-SearchBot, and PerplexityBot receive the sponsored markdown, while Google's standard search crawler gets the same HTML served to ordinary browsers. GPTBot and ChatGPT-User, which OpenAI uses for model training and user-initiated retrieval respectively, return HTTP 406 errors in Schmalbach's tests, while OAI-SearchBot receives the markdown version used to support ChatGPT Search. These AI ad-infused markdown versions of stories should be viewable by humans who change the User-Agent header sent as part of HTTP requests, which is how Schmalbach said he spotted them. The Register tested several articles from Time to see for ourselves, and we were able to replicate the results using the crawler simulator provided by search engine optimization firm Encited. When viewed as ClaudeBot, Time’s Best Inventions of 2025 list includes the Ally Bank ad, as do stories about Time’s top content creators of 2026, as does a gallery of photos from a creators' party held in New York recently to honor the list. The markdown ads don’t appear on newsier stories that we checked, suggesting the ads may be part of more evergreen content rather than articles with a shorter shelf life. As Schmalbach points out in his writeup, this could be “the first clear look at what the web starts to become when the main audience is AI models,” which is a threshold we’ve already crossed. Why advertise to humans when you can make AI do it for you? It’s not clear whether Time’s strategy of advertising to AI crawlers has been adopted by other publishers, but it’d be a shock if others weren’t at least considering it. As noted above, AI crawler traffic has already begun to surpass human visits to webpages, making it potentially more lucrative for advertisers to target those crawlers than the humans who made them. With half of the US consumers surveyed saying they use AI to search the web, the potential to reach consumers by influencing AI responses is incredibly powerful, and that is what Time and its advertising partner Mobian appear to be banking on. When we asked the publication to verify Schmalbach’s report and the results we found, they didn’t answer questions, instead pointing us to a story in media and marketing outlet Digiday from last week that verified not only what we saw but our fears about the scheme, too. Per Digiday, Time started converting its web pages to markdown in order to make them more appealing to AI crawlers in June, and a partnership with Mobian was included in that rework. Time COO Mark Howard told Digiday that the publication’s sales team is pitching agent ads to brands that have also converted their websites to markdown, as it suggests those brands are thinking about reaching AI bots. It’s a quote from Mobian CEO and co-founder Jonah Goodhart that makes the objective clear, however: The company's aim is to shape what AI assistants say about the brands paying for the service. “When you influence ChatGPT, you’re influencing potentially all of ChatGPT. If ChatGPT changes what it says about [a brand], it’s massive and it’s more than any one campaign could ever do,” Goodhart told Digiday. “With a human you influence one person.” So far, the pair confirmed to Digiday that Ally Bank and the Project Management Institute were the first brands trying to influence AI search results, matching the ads Schmalbach and The Register found in AI crawler versions of Time's pages. Time and Mobian told Digiday that the partnership marks the first time a publisher has served ads directly targeting AI crawlers, but that “more are already being lined up.” In other words, you soon may not be able to tell which results in an AI powered search are legitimately being surfaced because of being good products, and which are just gaming the system. It’s the early days of the SEO race all over again, only this time fooling the indexers is as easy as handing any old idiot the same pamphlet over and over again until its content becomes truth. ®