An anonymous reader quotes a report from Ars Technica: Recently, AI researcher Simon Willison wanted to add up his charges from using a cloud service, but the payment values and dates he needed were scattered among a dozen separate emails. Inputting them manually would have been tedious, so he turned to a technique he calls "video scraping," which involves feeding a screen recording video into an AI model, similar to ChatGPT, for data extraction purposes. What he discovered seems simple on its surface, but the quality of the result has deeper implications for the future of AI assistants, which may soon be able to see and interact with what we're doing on our computer screens.

"The other day I found myself needing to add up some numeric values that were scattered across twelve different emails," Willison wrote in a detailed post on his blog. He recorded a 35-second video scrolling through the relevant emails, then fed that video into Google's AI Studio tool, which allows people to experiment with several versions of Google's Gemini 1.5 Pro and Gemini 1.5 Flash AI models. Willison then asked Gemini to pull the price data from the video and arrange it into a special data format called JSON (JavaScript Object Notation) that included dates and dollar amounts. The AI model successfully extracted the data, which Willison then formatted as CSV (comma-separated values) table for spreadsheet use. After double-checking for errors as part of his experiment, the accuracy of the results -- and what the video analysis cost to run -- surprised him.

"The cost [of running the video model] is so low that I had to re-run my calculations three times to make sure I hadn't made a mistake," he wrote. Willison says the entire video analysis process ostensibly cost less than one-tenth of a cent, using just 11,018 tokens on the Gemini 1.5 Flash 002 model. In the end, he actually paid nothing because Google AI Studio is currently free for some types of use.

An anonymous reader shares a report: Every year for the past seven, Nathan Benaich, the founder and solo general partner at the early-stage AI investment firm Air Street Capital, has produced a magisterial "State of AI" report. Benaich and his collaborators marshal an impressive array of data to provide a great snapshot of the technology's evolving capabilities, the landscape of companies developing it, a survey of how AI is being deployed, and a critical examination of the challenges still facing the field.

One of the big takeaways from this year's report, which was published late last week, is that OpenAI's lead over other AI labs has largely eroded. Anthropic's Claude 3.5 Sonnet, Google's Gemini 1.5, X's Grok 2, and even Meta's open-source Llama 3.1 405 B model have equaled, or narrowly surpassed on some benchmarks, OpenAI's GPT-4o.ââBut, on the other hand, OpenAI still retains an edge for the moment on reasoning tasks with the release of its o1 "Strawberry" model -- which Air Street's report rightly characterized as a weird mix of incredibly strong logical abilities for some tasks, and surprisingly weak ones for others.

Another big takeaway, Benaich told me, is the extent to which the cost of using a trained AI model -- an activity known as "inference" -- is falling rapidly. There are several reasons for this. One is linked to that first big takeaway: With models less differentiated from one another on capabilities and performance, companies are forced to compete on price.ââAnother reason is that engineers for companies such as OpenAI and Anthropic -- and their hyperscaler partners Microsoft and AWS, respectively -- are discovering ways to optimize how the largest models run on big GPU clusters. The cost of outputs from OpenAI's GPT-4o today is 100-times less per token (which is about equivalent to 1.5 words) than it was for GPT-4 when that model debuted in March 2023. Google's Gemini 1.5 Pro now costs 76% less per output token than it did when that model was launched in February 2024.â

An anonymous reader shares a report: At the end of September, Kaspersky forcibly uninstalled and replaced itself with a new antivirus called UltraAV on the computers of around a million Americans, many of whom were surprised and aghast that they were not asked to give their consent for the change. The move was the end result of the U.S. government ban on all sales of Kaspersky software in the country and -- at least in theory -- marked the end of Kaspersky in America.

But not everyone in the U.S. has given up on the Russian-made antivirus. Some Americans have found ways to get around the ban and are still using Kaspersky's antivirus, TechCrunch has learned. Several people who live in the U.S. said in posts on Reddit that they are holding out as Kaspersky customers. When TechCrunch asked them about their motivations, their reasons range from being skeptical of the reasons behind the ban, or having paid for the product already, to simply preferring the product over its rivals.

The FIDO Alliance is developing new specifications to enable secure transfer of passkeys between different password managers and platforms. Announced this week, the initiative is the result of collaboration among members of the FIDO Alliance's Credential Provider Special Interest Group, including Apple, Google, Microsoft, 1Password, Bitwarden, Dashlane, and others. From a report: Passkeys are an industry standard developed by the FIDO Alliance and the World Wide Web Consortium, and were integrated into Apple's ecosystem with iOS 16, iPadOS 16.1, and macOS Ventura. They offer a more secure and convenient alternative to traditional passwords, allowing users to sign in to apps and websites in the same way they unlock their devices: With a fingerprint, a face scan, or a passcode.

Passkeys are also resistant to online attacks like phishing, making them more secure than things like SMS one-time codes. The draft specifications, called Credential Exchange Protocol (CXP) and Credential Exchange Format (CXF), will standardize the secure transfer of credentials across different providers. This addresses a current limitation where passkeys are often tied to specific ecosystems or password managers. Further reading: Passwords Have Problems, But Passkeys have more.

An anonymous reader shares a report: From the ground, northeastern Norway might look like fjord country, peppered with neat red houses and dissected by snowmobile tours through the winter. But for pilots flying above, the region has become a danger zone for GPS jamming. The jamming in the region of Finnmark is so constant, Norwegian authorities decided last month they would no longer log when and where it happens -- accepting these disturbance signals as the new normal.

Nicolai Gerrard, senior engineer at NKOM, the country's communications authority, says his organization no longer counts the jamming incidents. "It has unfortunately developed into an unwanted normal situation that should not be there. Therefore, the [Norwegian authority in charge of the airports] are not interested in continuous updates on something that is happening all the time." Pilots meanwhile, still have to adapt, usually when they are above 6,000 feet in the air. "We experience this almost every day," says Odd Thomassen, a captain and senior safety adviser at the Norwegian airline Wideroe. He claims jamming typically lasts between six and eight minutes at a time.

With ransomware attacks surging and 2024 on track to be one of the worst years on record, U.S. officials are seeking ways to counter the threat, in some cases, urging a new approach to ransom payments. From a report: Ann Neuberger, U.S. deputy national security adviser for cyber and emerging technologies, wrote in a recent Financial Times opinion piece, that insurance policies -- especially those covering ransomware payment reimbursements -- are fueling the very same criminal ecosystems they seek to mitigate. "This is a troubling practice that must end," she wrote, advocating for stricter cybersecurity requirements as a condition for coverage to discourage ransom payments.

Zeroing in on cyber insurance as a key area for reform comes as the U.S. government scrambles to find ways to disrupt ransomware networks. According to the latest report by the Office of the Director of National Intelligence, by mid-2024 more than 2,300 incidents already had been recorded -- nearly half targeting U.S. organizations -- suggesting that 2024 could exceed the 4,506 attacks recorded globally in 2023. Yet even as policymakers scrutinize insurance practices and explore broader measures to disrupt ransomware operations, businesses are still left to grapple with the immediate question when they are under attack: Pay the ransom and potentially incentivize future attacks or refuse and risk further damage.

For many organizations, deciding whether to pay a ransom is a difficult and urgent decision. "In 2024, I attended a briefing by the FBI where they continued to advise against paying a ransom," said Paul Underwood, vice president of security at IT services company Neovera. "However, after making that statement, they said that they understand that it's a business decision and that when companies make that decision, it is taking into account many more factors than just ethics and good business practices. Even the FBI understood that businesses need to do whatever it takes to get back to operations," Underwood said.

An anonymous reader shares a report: Byju Raveendran, the founder of the embattled edtech group Byju's, acknowledged on Thursday afternoon that he made mistakes, mistimed the market, overestimated growth potential and that his startup, once valued at $22 billion, is now effectively worth "zero."

Speaking to a group of journalists, Raveendran said the company's aggressive acquisition of more than two dozen startups to expand into new markets proved fatal when financing dried up in 2022. Byju's was planning to go public in early 2022 with several investment bankers giving the firm valuation as high as $50 billion, TechCrunch reported earlier.

He alleged that many of his more than 100 investors had urged him to pursue aggressive expansion into as many as 40 markets. But, he added, those very investors got cold feet when global markets tumbled following Russia's invasion of Ukraine, sending the venture capital market into a downward spiral.

WP Engine has filed a motion for a preliminary injunction against Automattic and its CEO Matt Mullenweg, seeking to halt their public campaign and regain access to WordPress resources. The hosting platform claims it's suffering "immediate irreparable harm," including a 14% spike in cancellation requests following Mullenweg's criticism.

WP Engine alleges the dispute has created anxiety among developers and increased security risks for the WordPress community. The legal action comes after Automattic accused WP Engine of trademark infringement, leading to exchanged cease-and-desist orders and a lawsuit. Last week, the WordPress.org project, led by Mullenweg, took control of WP Engine's Advanced Custom Fields plugin, redirecting users to a forked version.

India is expected to limit imports of laptops, tablets and personal computers after January, Reuters reported Friday citing government sources, a move to push companies such as Apple to increase domestic manufacturing. From the report: This plan, if implemented, could disrupt an industry worth $8 billion to $10 billion and reshape the dynamics of the IT hardware market in India, which is heavily reliant on imports. A similar plan to restrict imports was withdrawn last year following backlash from companies and lobbying from the United States. India has since monitored imports under a system set to expire this year and has asked firms to seek fresh approvals for imports next year. The government feels it has given the industry enough time to adapt, said the sources, who did not want to be identified as discussions are private.

Salesforce founder and chief executive Marc Benioff has doubled down on his criticism of Microsoft's Copilot, the AI-powered tool that can write Word documents, create PowerPoint presentations, analyze Excel spreadsheets and even reply to emails through Outlook. In a post on X, he writes: When you look at how Copilot has been delivered to customers, it's disappointing. It just doesn't work, and it doesn't deliver any level of accuracy. Gartner says it's spilling data everywhere, and customers are left cleaning up the mess.

To add insult to injury, customers are then told to build their own custom LLMs. I have yet to find anyone who's had a transformational experience with Microsoft Copilot or the pursuit of training and retraining custom LLMs. Copilot is more like Clippy 2.0.

An anonymous reader quotes a report from TechCrunch: Stripe is in talks to acquire stablecoin platform Bridge for a whopping $1 billion, according to Forbes (paypalled). The talks are reportedly in advanced stages, although nothing has been finalized. Bridge, co-founded by Coinbase alumni Zach Abrams and Sean Yu, has built an API that helps companies accept stablecoins. The pair raised $58 million from investors like Index Ventures and Sequoia Capital, according to PitchBook. If the deal with Stripe goes through, it would be a huge jump from Bridge's $200 million valuation, as well as being Stripe's largest acquisition to date.

The blockchain-based identity verification company founded by Sam Altman is now called "World." It also unveiled a new version of the "Orb" biometric devices the company uses to scan users' eyes. CoinTelegraph reports: World, as it's now known, also revealed a slew of other updates including a new version of its Orb biometric scanning devices, new options for identity verification and partnership integrations with popular apps including FaceTime, WhatsApp, and Zoom. [...] The new Orb, powered by Nvidia hardware, will be more efficient and "five times" more powerful than its predecessor with a smaller footprint and fewer parts. The company also said the new Orb would eventually be available in self-service kiosks in some markets.

World also announced that users will soon be able to verify their identity through methods other than the firm's Orb hardware. Through a program called World ID Credentials, the company says users with NFC-enabled government issued passports will allow them to verify their identity on the World app. Another major announcement came in the form of World ID Deep Face, a service the company claims has "solved deepfakes." According to the company, its software can be implemented into just about any app where video can be uploaded or streamed to determine whether videos featuring verified persons are real or have been faked using AI. Finally, the company also announced that so far 15 million users have signed up for its World app service; among them, seven million are verified.

AWS CEO Matt Garman has harsh words for remote workers: return to the office or quit. TechCrunch: The Amazon executive recently told employees who don't like the new five-day in-person work policy that, "there are other companies around," presumably companies they can work for remotely, Reuters reported on Thursday. Amazon's top boss, Andy Jassy, told employees last month that there will be a full return-to-office starting in 2025, an increase from three days for roughly the last year.

An anonymous reader quotes a report from The Verge: Donald Trump said Apple CEO Tim Cook called him to discuss the billions of dollars that Apple has been fined in the European Union. Trump made the statement during his appearance on the PBD Podcast -- and said that he won't let the EU "take advantage" of US companies like Apple if reelected. "Two hours ago, three hours ago, he [Cook] called me," Trump said. "He said the European Union has just fined us $15 billion... Then on top of that, they got fined by the European Union another $2 billion." In March, the EU fined Apple around $2 billion after finding that Apple used its dominance to restrict music streaming apps from telling customers about cheaper subscription deals outside the App Store. The EU later won its fight to make Apple pay $14.4 billion in unpaid taxes.

"He [Cook] said something that was interesting," Trump said. "He said they're using that to run their enterprise, meaning Europe is their enterprise. "I said, 'That's a lot... But Tim, I got to get elected first, but I'm not going to let them take advantage of our companies -- that won't, you know, be happening.'" Trump has talked to several Big Tech executives over the past several months. "During an interview this week, Trump said he spoke with Google CEO Sundar Pichai to complain about all the 'bad stories' the search engine shows about him," notes The Verge. "Elon Musk recently spoke at a Trump rally in Pennsylvania, while Meta CEO Mark Zuckerberg called Trump over the summer 'a few times,' according to the former president."

During its MAX event yesterday, Adobe teased some experimental photo and video editing tools for PhotoShop and Premiere Pro. There are a total of nine features, which include being able to rotate vector images, produce sound effects from text descriptions, and generate images in various shapes and sizes. Engadget reports: [W]e'll start with Project Perfect Blend for PS, which improves natural blending and makes shadow casting more realistic, creating more lifelike images. Project Clean Machine removes photo flashes, fireworks and objects blocking the camera's view. One feature that stands out is Project In Motion, which lets users transform custom shape animations into video by entering a prompt, while Project Know How is a content authenticator tool that can search for a video file's source online. Project Turntable lets users rotate 2D vector art in 3D, thereby allowing the 2D vector art to face a direction of their choice. The generative AI model fills in any blanks to create presentable 3D vector art.

Another standout tool is Project Super Sonic, which generates sound effects via prompts or clicking on objects in a video. The latter method can create sounds without typing prompts into the generative AI model. Project Super Sonic seems helpful for people looking to design the sounds they want. Adobe is also working on Microsoft Copilot integration in Project Scenic. This tool creates 3D scene layouts using Copilot prompts, and the camera and objects in the layout can be tweaked. Project Remix A Lot leverages generative AI to create images in various shapes and sizes, all fully editable. In other words, users can "remix" creations into shapes they like, including unusual ones. Finally, we have Project Hi-Fi. With this tool, it's possible to transform sketches and concepts into high-quality images. These images can easily be dragged into PhotoShop for editing.