14 October 2024

TechCrunch: “Meta confirms it may train its AI on any image you ask Ray-Ban Meta AI to analyze”

In other words, the company is using its first consumer AI device to create a massive stockpile of data that could be used to create ever-more powerful generations of AI models. The only way to “opt out” is to simply not use Meta’s multimodal AI features in the first place.

The implications are concerning because Ray-Ban Meta users may not understand they’re giving Meta tons of images – perhaps showing the inside of their homes, loved ones, or personal files – to train its new AI models. Meta’s spokespeople tell me this is clear in the Ray-Ban Meta’s user interface, but the company’s executives either initially didn’t know or didn’t want to share these details with TechCrunch. We already knew Meta trains its Llama AI models on everything Americans post publicly on Instagram and Facebook. But now, Meta has expanded this definition of “publicly available data” to anything people look at through its smart glasses and ask its AI chatbot to analyze.

Maxwell Zeff

I’ve been vaguely following the news around these new smart glasses from Meta to see how they would fare compared to Google’s failed attempt with Glass a decade ago, and Apple’s massive VR helmet. This reporting underscores what some remarked repeatedly about ad-driven companies like Google and Facebook: their products are built for the core purpose of data collection for ad targeting. The privacy implications are as egregious as with Google Glass: everything in the visual field of these glasses will likely end up in a facial recognition database and might get exposed publicly – in fact, some college students have already hacked the Ray-Ban Meta glasses to reveal the name, address, and phone number of anyone they look at.

24 September 2024

ABC News: “Facebook admits to scraping every Australian adult user’s public photos and posts to train AI, with no opt-out option”

But that was quickly challenged by Greens senator David Shoebridge.

Shoebridge: The truth of the matter is that unless you have consciously set those posts to private since 2007, Meta has just decided that you will scrape all of the photos and all of the texts from every public post on Instagram or Facebook since 2007, unless there was a conscious decision to set them on private. That’s the reality, isn’t it?

Claybaugh: Correct.

Ms Claybaugh added that accounts of people under 18 were not scraped, but when asked by Senator Sheldon whether public photos of his own children on his account would be scraped, Ms Claybaugh acknowledged they would.

Jake Evans

Hardly surprising coming from Facebook – or any other American tech giant for that matter; the latest example being LinkedIn, which has covertly opted its users in to supply data for AI training without even bothering to update their terms of service.

This is why GDPR is so important in the EU, and why, despite all its critics and lackluster enforcement, the core of the legislation is sound and future proof. In simple terms, companies need to have legitimate reason, or consent, to use consumers’ data for new purposes. Training AI models is clearly a new purpose that doesn’t precisely have a legitimate reason, so companies should in theory ask for distinct consent before using the data.

14 September 2024

Financial Times: “How Lidl accidentally took on the big guns of cloud computing”

Starting with a system built for internal use in 2021, Lidl owner Schwarz Group now offers cloud computing and cyber security services to corporate customers.

Its IT unit, Schwarz Digits — which became a standalone operating division in 2023 — has signed up clients including Germany’s biggest software group SAP, the country’s most successful football club Bayern Munich and the port of Hamburg. Last year, the unit generated €1.9bn in annual sales and it employs 7,500 staff.

We did not start with a commercial motivation in mind but just wanted to address our own needs, Christian Müller, co-chief executive of Schwarz Digits told the Financial Times in a rare interview. We’re on a very steep growth path.

A main selling point of its service is that all client data is processed and stored exclusively in Germany and Austria, which have stringent privacy and data protection laws.

Olaf Storbeck

Exciting development for a European business to build its own cloud computing system from the ground up – and striking how closely this origin story resembles Amazon’s AWS. I would say this is a good example of how regulation can foster business opportunities. It would probably have been faster and cheaper for Schwarz Group to contract an existing US cloud provider, but since their products are not prioritizing privacy, the more compliant answer was to build an internal solution.

26 August 2024

Variety: “‘Megalopolis’ Marketing Consultant dropped over AI Quotes in Trailer”

Lionsgate has parted ways with Eddie Egan, the marketing consultant who came up with the “Megalopolis” trailer that included fake quotes from famous film critics.

The studio pulled the trailer on Wednesday, after it was pointed out that the quotes trashing Francis Ford Coppola’s previous work did not actually appear in the critics’ reviews, and were in fact made up.

Sources tell Variety it was not Lionsgate or Egan’s intention to fabricate quotes, but was an error in properly vetting and fact-checking the phrases provided by the consultant. The intention of the trailer was to demonstrate that Coppola’s revered work, much like “Megalopolis”, has been met with criticism. It appears that AI was used to generate the false quotes from the critics.

Gene Maddaus & Katcy Stephan

Perfect illustration of LLMs as unreliable creations, making them unfit tools for a dependable business environment in their current form. As many have pointed out, the biggest threats from current AI models are not human extinction, but the proliferation of falsehood, the dilution of truth, and making it trivial for people to get by with being sloppy and incompetent. This high-profile case was easy to spot, but who knows how many smaller projects are filled with made-up ‘facts’ and fabricated quotes.

13 August 2024

Mashable: “Elon Musk’s X lets users sort replies to find more relevant comments”

The latter two reply sorting options are pretty straightforward. “Most recent” shows replies in chronological order from newest to oldest. “Most liked” shows the replies with most likes first. Blue check accounts appear to have completely lost any advantage that the paid subscription provided them when selecting either of these two menu options.

It’s unclear exactly how X is determining how to sort posts via the “most relevant” option. However, it appears to be the same sorting method as the previous default. Blue checks do still appear to be prioritized in this view.

Any user can change the reply sorting options on any post that they view, not just their own posts.

Matt Binder

I’ve noticed this change a few days before this article, first on the Twitter web site, later in the Android app as well. Slightly impressive that the company is now shipping features simultaneously on the web and in its apps, whereas for years the standard for most tech giants was to test and launch features on iOS first, with the web and Android distant seconds. A lack of polish is however manifest in small details like different labels for the choices on the web and in apps (in the Android app, ‘Most relevant’ is called ‘Trending replies’). Unfortunately, this new feature is per post rather than saved globally on your account, meaning you can’t (yet?) choose your preferred reply sorting setting and have it automatically applied on each reply section, instead you need to switch modes on each new tweet.

11 August 2024

The Verge: “Why Google decided now’s the time to move on from Chromecast”

The Chromecast is going away after more than a decade in Google’s hardware portfolio, with the company phasing it out to make room for the new Google TV Streamer. In 2013, the timing was perfect for the Chromecast’s success. Most TVs of that era had rudimentary entertainment apps that were often slow, so beaming content from your phone to a TV made a lot of sense.

But times have changed. With the Chromecast with Google TV in 2020, Google pivoted in the opposite direction with a much more traditional, lean-back entertainment experience. Casting took a back seat to carousels of content recommendations, and native apps returned.


And yet, even right after the announcement, some of our commenters were frustrated that Google had settled for just a 22 percent boost in CPU speed. There’s a good chance this thing will still benchmark underneath the fairly ancient Shield. Why not just drop a Tensor chip into it?

It comes down to cost and keeping the Streamer at a price that’s acceptable for average consumers. We don’t know if there’s functionality that would actually convince people to buy pricier price points than this, but the market is generally telling us right now people are probably not ready for it, Govil-Pai said.

Chris Welch

I have been quite skeptical initially of the Chromecast, but after I purchased one to stream HBO to my non-so-smart former TV I have grown to appreciate its utility and simplicity. While I understand the reasoning behind the decision to discontinue it (nowadays TVs ship with Google TV and Chromecast protocols built-in; I myself have abandoned my old Chromecast when I upgraded the TV last year), it still feels a bit sad and disappointing.

10 August 2024

Wired: “Perplexity is a Bullshit Machine”

On June 6, Forbes published an investigative report about how former Google CEO Eric Schmidt’s new venture is recruiting heavily and testing AI-powered drones with potential military applications. (Forbes reported that Schmidt declined to comment.) The next day, John Paczkowski, an editor for Forbes, posted on X to note that Perplexity had essentially republished the sum and substance of the scoop. (It rips off most of our reporting, he wrote. It cites us, and a few that reblogged us, as sources in the most easily ignored way possible.)

That day, Srinivas thanked Paczkowski, noting that the specific product feature that had reproduced Forbes’ exclusive reporting had rough edges and agreeing that sources should be cited more prominently. Three days later, Srinivas boastedinaccurately, it turned out—that Perplexity was Forbes’ second-biggest source of referral traffic. (WIRED’s own records show that Perplexity sent 1,265 referrals to WIRED.com in May, an insignificant amount in the context of the site’s overall traffic. The article to which the most traffic was referred got 17 views.) We have been working on new publisher engagement products and ways to align long-term incentives with media companies that will be announced soon, he wrote. Stay tuned!


In theory, Perplexity’s chatbot shouldn’t be able to summarize WIRED articles, because our engineers have blocked its crawler via our robots.txt file since earlier this year. This file instructs web crawlers on which parts of the site to avoid, and Perplexity claims to respect the robots.txt standard. WIRED’s analysis found that in practice, though, prompting the chatbot with the headline of a WIRED article or a question based on one will usually produce a summary appearing to recapitulate the article in detail.

Dhruv Mehrotra & Tim Marchman

Another looming issue over the blooming LLM industry: plagiarism and copyright violations. In their hunger for information to train models, AI companies have repeatedly made dubious decisions that have resulted in a number of damning situations. The most prominent was probably OpenAI releasing a synthetic voice which sounded very similar to Scarlett Johansson; Wired reporters were able to download paywalled articles from publishers like The New York Times and The Atlantic through Quora’s Assistant bot; and the practice is not limited to startups either, as another investigation found that Apple, Nvidia, and Salesforce used thousands of YouTube videos to train AI.