This is the most well-articulated discussion I’ve seen so far, of the ethical questions of generative AI.

I heard that they require plaintext data to work. What are the other factors to this?

Oh, the irony ...

Meta will begin tracking the mouse movements, clicks, and keystrokes of its US employees to generate high-quality training data for future AI agents, Reuters reports.

The news organization cites internal memos posted by the Meta Superintelligence Labs team in reporting on the new Model Capability Initiative employee-tracking software. That software will operate on specific work-related apps and websites and also make use of periodic screenshots to provide context for the AI training, according to the memo.

“This is where all Meta employees can help our models get better simply by doing their daily work,” the memo reads, in part, Reuters reports.

Meta spokesperson Andy Stone told Reuters that the collected training data will help Meta’s AI agents with tasks that it sometimes struggles with, including “things like mouse movements, clicking buttons, and navigating dropdown menus.”

“If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how we actually use them,” Stone said, adding that the collected data would not be used to evaluate employees.

Sure it won't. I find it hard to believe this isn't just admitting to systems that were already in place.

There were clues from the start that it was too good to be true. A headhunter emailed me with a job prospect – a journalist role with “a leading US technology and markets editorial team”. The opportunity, she said, was part of a confidential expansion and hadn’t been publicly posted.

My spidey-sense was tingling, but the timing was auspicious. I was on the lookout for new work as my maternity leave was coming to an end. Initially, the email seemed legitimate. When I Googled the sender, I found a headhunter with the same name and profile picture on LinkedIn, and the message was clearly tailored to me: It referenced several roles I’d previously held and identified my specific areas of expertise. “Your focus on the real-world impacts of AI, digital culture and the gig economy aligns perfectly with an internal, high-priority mandate I’m managing,” the headhunter wrote.

I emailed back. The headhunter asked me to send over my CV, along with my salary expectations, preferred work structure (remote, hybrid, or on-site), and geographic flexibility. In return, she shared a more detailed job description. The role was, indeed, perfect for me. Too perfect – as if someone had put my CV into ChatGPT and asked it to create a job description based directly on my experience. It was located in the city in which I live and offered a hybrid working arrangement, just as I’d requested. The biggest tell: I’d been ambitious with my salary suggestion, but this was offering significantly more.

By this point I was fairly sure I was being taken for a ride, but I still couldn’t figure out the scam. I found myself trying to justify the anomalies. It’s an American company, and salaries are generally higher there, aren’t they? I asked about next steps. Then the headhunter gave me feedback. My CV undersold my leadership skills, she said; it needed refining. If I liked, she could connect me with a specialist who would make my profile more compelling. They would discuss pricing directly with me.

I fell for a "resume help" scam a couple of years ago via LinkedIn. I don't even use the site anymore.

Internal documents reveal that Microsoft plans to temporarily suspend individual account signups to its GitHub Copilot coding product, as it transitions from requests (single interactions with Copilot) towards token-based billing.

The documents reveal that the weekly cost of running Github Copilot has doubled since the start of the year.

Microsoft also intends to tighten the rate limits on its individual and business accounts, and to remove access to certain models for those with the cheapest subscriptions.

archive link

cross-posted from: https://lazysoci.al/post/46451012

A ClickFix campaign targeting macOS users delivers an AppleScript-based infostealer that collects credentials and live session cookies from 14 browsers, 16 cryptocurrency wallets, and more than 200 extensions.

Netskope Threat Labs researcher Jan Michael Alcantara told The Register the team initially observed the campaign last month, and has seen similar instances as recently as last week.

ClickFix is a super popular social engineering tactic used to trick people into executing malicious commands on their own computers, usually by clicking a fake computer problem fix or CAPTCHA prompt.

While the researchers don't know who the cookie thief is, they note the malware can infect both Windows and macOS machines - Netskope previously warned about the Windows-focused attacks - by using a client-side JavaScript to filter victims by user-agent, ignoring mobile devices and directing desktop users to either a Windows or macOS-specific payload.

Apple announced on Monday that it had named a replacement for Tim Cook as CEO, with head of hardware engineering John Ternus succeeding him on 1 September. Cook will stay at the company in the role of executive chair.

“It has been the greatest privilege of my life to be the CEO of Apple and to have been trusted to lead such an extraordinary company. I love Apple with all of my being,” Cook said in a press release.

Cook, who succeeded Apple co-founder Steve Jobs, has been CEO since 2011. He has overseen the global expansion of the company and its steady series of new, updated devices, though never attained the same tech visionary status as Jobs.

Cook’s tenure as CEO has marked a lucrative period of expansion for Apple as it entrenched its products in society and sought out new markets, in particular the iPhone. Apple reported earlier this year it had its best ever quarter for iPhone sales, driven by renewed demand in China. The company’s market capitalization grew from around $350bn at the start of Cook’s time to over $4tn today.

cross-posted from: https://lazysoci.al/post/46451015

In the day that I have had access, I gave the new model a wide range of tasks.

• A friend asked me to make a memorial image of her recently deceased cat along with two favorite toys. It crafted an image that looked like a highly personalized sympathy card.

• It elegantly took two photos from my wedding and made it appear as if they were in an old-style photo album with photo corners.

• My colleagues suggested a poster for a fictional event. I decided to create a Mike Allen look-alike contest in Washington Square Park this Sunday. (Of course, it's only fictional if no one shows up.)

It also made a handy infographic making "the case against candy corn" which I used unsuccessfully to convince two colleagues that the treat, which is neither candy nor corn, is also not good.

The full pages it designs are scary good. I'd go so far as to say this is definitely a shot across the bow for design work.

If you're used to absurd lettering and poor design decisions, the output included in the story suggests otherwise.

Hundreds of previously redacted records reveal how Amazon has put pressure on independent sellers using its platform into raising their prices on the sites of competitors such as Walmart and Target, so that Amazon can appear to have lower prices, California authorities allege.

The global conglomerate became concerned even if a competitor was selling an item for as little as a penny less, according to one segment of the newly unredacted evidence.

The documents – which have never previously been reported on – include internal emails, deposition testimony and confidential corporate presentations that the California attorney general, Rob Bonta, obtained as part of a civil case his office launched in 2022 accusing Amazon of large-scale price-fixing.

The Guardian obtained and reviewed the cache of evidence, which has been filed in San Francisco county superior court but has not yet become publicly available. Within the documents, lawyers for the state of California have unmasked key details, paragraphs and sometimes whole pages that had previously been blacked out. A judge permitted some redactions to remain at Amazon’s request.

OnlyFans, the UK adult video platform, is in talks to sell a minority stake to a US investor that will value the business at more than $3bn (£2.2bn).

The London-based company is in advanced talks to sell a stake of less than 20% to the San Francisco-based investment firm Architect Capital, according to the Financial Times. Sources familiar with the process confirmed the talks to the Guardian.

OnlyFans has decided that offloading a minority stake is the best guarantee of stability for a business dealing with the death of its owner, Leonid Radvinsky. Radvinsky, a Ukrainian-American billionaire, died of cancer last month at the age of 43.

It is understood that OnlyFans is interested in a deal with Architect because the firm has expertise in the financial services sector, reflecting the UK company’s interest in offering banking products to its creators, who have struggled to access such services owing to the nature of their work.

Google began rolling out “personal intelligence” in Gemini early this year, giving AI subscribers the option of a more customized experience when using the company’s chatbot. Today, it’s using personal intelligence to tie its image-generation model to Google Photos. If you opt in, generated images will have access to your photos and associated labels to simplify prompts and produce more accurate AI images.

This change essentially streamlines an existing workflow. Google’s Nano Banana 2 is among the best AI image generators available, and it was already possible to feed it images of yourself or others to use as context for creating new AI content. Adding personal intelligence to the mix makes that process smoother by turning the image bot loose on the content of your photos, if indeed that’s something you want to do.

It is generally true that adding more personal data to an AI prompt results in a better output. Google offers a few examples of how connecting Nano Banana to Photos can help in this way. You won’t have to pack as much context into your prompts—you can just refer to “my family” or “my dog” to let the robot find useful images in your Photos library.

Just what I need. Family photos that never happened. "OK, Google, show me a Christmas photo where my dad actually went out for a pack of smokes and immediately returned."