IT News Review by Control F5 Software: Noted AI researcher launches controversial startup that wants to replace all human employees

A well-known AI researcher launches controversial startup that wants to replace all human employees

Every once in a while, a startup pops up around Silicon Valley with a mission so "absurd" that it's difficult to discern whether the startup is real or just satire.

That's the case with Mechanize, a startup whose founder - and the nonprofit AI research organization he founded, called Epoch - is being criticized on X after announcing its launch.

Mechanize was launched Thursday in a post on X by its founder, renowned AI researcher Tamay Besiroglu. The startup's goal, Besiroglu wrote, is "complete automation of any job" and "complete automation of the economy."

A new battery technology produces energy and cleans wastewater at the same time

Researchers at the Indian Institute of Science (IISc) have introduced an innovation that could transform both energy storage and industrial pollution treatment. Published in the journal ACS Applied Materials & Interfaces, the new technique uses zinc-air battery technology to produce hydrogen peroxide (H₂O₂) in a cleaner, cheaper and potentially scalable way - all while combating toxic dyes in the textile industry.

The study is at the intersection of energy innovation, green chemistry and saving the environment. It paves the way for multifunctional energy systems, capable not only of storing electricity but also of performing useful chemical reactions as part of their operation.

An Italian newspaper unleashes AI and admires its irony

Artificial intelligence can write an excellent book review and is good at irony, but it will not replace quality journalism, said the editor of an Italian newspaper at the forefront of AI experimentation.

In a world first, Il Foglio published for a month a four-page supplement written entirely by AI and sold with the regular newspaper.

The experiment, which has just ended, was a great success and boosted sales, said editor Claudio Cerasa, adding that his paper will now publish a separate section written by AI once a week.

FDA-approved brain implant for Precision Neuroscience, a rival to Musk's Neuralink

Neurotech startup Precision Neuroscience announced Thursday that a key component of its brain implant system has been approved by the U.S. Food and Drug Administration, a major victory for the company founded four years ago.

Precision is building a brain-computer interface, or BCI, which is a system that decodes neural signals and translates them into commands for external technologies. The company's BCI will initially be used to help patients with severe paralysis regain functions such as speech and movement, according to its website.

Only part of the Precision system was approved Thursday by the FDA.

Google One AI Premium is free for students through spring 2026

Google is the latest AI service provider to woo higher education users. US students can sign up for Google's One AI Premium plan for free, avoiding the usual $20 monthly fee until June 30, 2026

According to Google spokesperson Alex Joseph, applicants will need to sign up before the offer expires and will need a valid .edu email address for verification. He told The Verge that students will receive an email toward the end of the plan, "so you have plenty of time to cancel."

New emails showed how Meta has struggled to keep Facebook culturally relevant

In a series of emails from April 2022, which were introduced into evidence during a trial, Facebook founder Mark Zuckerberg discussed concerns that Facebook's "Friends" structure and feed are outdated as all other major platforms focus on "Follow."

He considered scrapping the feed and even suggested deleting all Facebook friends and starting from scratch.

China will bet on artificial intelligence in education reform

China will integrate artificial intelligence (AI) applications into teaching efforts, school textbooks and curriculum as part of its education reform process, authorities announced in an official document released Wednesday.

The measure, which targets students and teachers in primary, secondary and higher education, comes as the world's second-largest economy seeks to spur innovation and find new sources of growth.

Promoting artificial intelligence would help "nurture the basic skills of teachers and students" and build "core competitiveness of innovative talents", the education ministry said.

Privacy data collection makes mobile apps a major target for hackers

Mobile apps are quietly increasingly attracting the attention of malicious people. They contain a wealth of private information about their users. In the iOS universe alone, 82.78%, or roughly 1.55 million apps, track users' private data, according to Exploding Topics, a trend-monitoring service.

Mobile apps have also proven particularly vulnerable to cybercriminals. "Invisible" entry and exit points in mobile apps can be compromised before traditional security tools detect a breach. These points include API calls, background synchronization and push notifications.

Perplexity is reportedly the key to Motorola's next Razr

Perplexity's AI voice assistant Perplexity will play a significant role in the upcoming Motorola Razr, according to Bloomberg. The news comes after Motorola posted a teaser video of the Razr on social media last week, in which the foldable device animates itself by forming the word AI.

Perplexity is also working with T-Mobile's parent company T-Mobile on a new "AI phone" with agents that could handle tasks such as booking flights without the user having to interact with apps.

Strava acquires hugely popular Runna app

Strava, an app for active people, announced that it has entered into a definitive agreement to acquire Runna, a UK-based technology company that develops personalized training and coaching plans for running.

The acquisition unites the world's largest fitness community with a leading app in the highly competitive field of running training.

The first foldable e-reader is more compact than a paperback book

Foldable smartphones with OLED displays have been around for more than six years, but Readmoo, a China-based company, is the first to launch a consumer e-reader with a foldable E-Ink display. Although the price and release date for the new mooInk V are still a secret, Readmoo says it has been working with E-Ink for nine years to develop a screen durable enough to withstand more than 200,000 folds.

When opened, the mooInk V reveals an 8-inch touchscreen built on E-Ink's advanced Gallery 3 technology. Unlike the Kindle Colorsoft's Kaleido 3 display, which uses a color filter and is limited to 4,096 shades, Gallery 3 uses colored ink particles to deliver more than 50,000 shades of color at 300 pixels per inch. This technology is currently only found on a few devices, such as the Remarkable Paper Pro.

Wikipedia offers an AI-optimized dataset to reduce bot scraping

In an effort to reduce the pressure on servers caused by automated scraping with AI, Wikipedia is proactively offering a new dataset tailored to the development of artificial intelligence. The Wikimedia Foundation has partnered with Kaggle, the Google-owned data science platform, to launch a beta version of Wikipedia's structured content in English and French.

Designed with machine learning workflows in mind, the dataset provides developers with clean, readable data ideal for training, fine-tuning, benchmarking, and analytics. As of April 15, the dataset includes research summaries, short descriptions, data from fact boxes, links to images, and article sections - excluding quotes and non-text elements such as audio files.

Experts say Google's new Gemini 2.5 Pro report lacks crucial safety information

Weeks after the launch of the Gemini 2.5 Pro, Google's most advanced AI model to date, the company published a technical report on internal safety assessments. But experts say the report provides limited details, leaving key questions about potential risks unanswered.

Technical reports are typically considered essential tools for transparency, providing both positive and critical information about the performance and safety of AI models. While many members of the AI community consider these reports to be bona fide contributions to independent oversight and research, Google's most recent publication falls short of expectations.

An AI support bot sparks outrage after inventing company policy

A recent incident involving AI-powered code editor Cursor has reignited concerns about the reliability of artificial intelligence in customer support roles. On Monday, a developer noticed that switching from one device to another while using Cursor unexpectedly disconnected it, a frustrating problem for programmers who rely on seamless workflows across multiple devices.

Looking for an answer, the user contacted support and received a response from an agent named "Sam," who explained that the behavior was due to a new policy that limited usage to one device per subscription, citing security reasons. The explanation seemed legitimate, except that it wasn't true. "Sam" was not a human, but an AI support bot, and no such policy existed.

OpenAI's new reasoning models have more hallucinations than the older ones

OpenAI's latest AI models, the o3 and o4-mini, are designed to be top-notch reasoning engines, but they have a significant drawback: they have more frequent hallucinations than several of the company's previous models.

Hallucinations, in which AI generates false or fabricated information, remain one of the most persistent challenges in the field. While previous models have generally shown gradual improvements in reducing hallucinations, the newest versions of OpenAI appear to buck this trend.

According to internal tests, o3 and o4-mini, categorized as reasoning models, perform worse in this respect than previous models such as o1, o1-mini and o3-mini, as well as more traditional models such as GPT-4o.

Google loses ad tech monopoly lawsuit

The US Department of Justice has won its antitrust lawsuit against Google, which accused the company of operating a monopoly in the advertising technology industry. The ruling, which marks Google's latest antitrust defeat since the search engine lawsuit, alleges that the tech giant's anticompetitive practices in two key markets "substantially harmed" publishers and internet users.

"Plaintiffs have proven that Google willfully engaged in a series of anticompetitive acts to acquire and maintain monopoly power in the markets for publisher ad servers and ad-sharing for display advertising on the open Web," writes U.S. District Judge Leonie Brinkema. "For more than a decade, Google has tied its publisher ad server and ad exchange through contractual policies and technological integration, which has allowed the company to establish and protect its monopoly power in these two markets."

HP accepts $4 million settlement following allegations of "false advertising" for PCs and keyboards

HP Inc. has agreed to pay a $4 million settlement to customers after being accused of "false advertising" for computers and peripherals on its Web site.

Earlier this month, Judge P. Casey Pitts of the U.S. District Court in the San Jose Division of the Northern District of California granted preliminary approval [PDF] of a settlement agreement on a class action complaint originally filed against HP on October 13, 2021. The complaint accused HP's website of displaying "misleading" original prices for various computers, mice, and keyboards that were higher than the products' recent and customary prices.

ABB to spin off world's second-largest robotics business

Swiss industrial group ABB announced plans on Thursday to spin off its robotics division, in the company's biggest reorganization since the sale of its power grids business to Japan's Hitachi in 2018.

The company is the world's second-largest maker of industrial robots after Japan's FANUC Corp and competes with rivals such as Yaskawa and Germany's Kuka.

It generated sales of $2.3 billion in 2024, equivalent to 7% of ABB's total, but has struggled in recent quarters amid falling demand from the automotive sector, a major buyer of robots.

Summary compiled using a monitoring feed provided by Control F5 Software.