Tech 16 July 2026

The AI Act gets serious on August 2 — here's what changes for anyone using AI in Portugal

On 2 August 2026 the AI Act's core obligations apply, including the rules for high-risk AI systems used in hiring, credit, education or essential services. What changes for companies in Portugal.

Straight answer: on 2 August 2026 the AI Act — the EU’s artificial intelligence regulation — enters its most demanding phase, as the obligations for high-risk systems and most transparency rules start to apply. For many companies in Portugal using AI in decisions about people, the warm-up period ends this summer.

The regulation has been in force since August 2024, but it was built to land in stages: outright bans first (manipulation, social scoring), then the rules for large general-purpose models, and now the heaviest layer.

What changes on 2 August 2026?

The obligations for high-risk AI systems on the regulation’s list kick in — those used in hiring and workforce management, access to education, credit scoring, insurance, essential public services or border control. Providers and deployers must ensure risk management, data quality, technical documentation, human oversight and logging. Transparency duties stack on top: chatbots must identify themselves as machines, and synthetic content such as deepfakes must be labelled. Member states are also required to have at least one AI regulatory sandbox running. The full official timeline is on the European Commission’s page.

What happens if a company ignores the AI Act?

The fines have a serious ceiling: up to 35 million euros or 7% of global turnover for the gravest breaches. Enforcement is split between national authorities and the European AI Office, which now has an independent scientific panel watching frontier models. Not everything lands yet: rules for AI embedded in regulated products, like medical devices, only arrive in 2027.

For Portuguese SMEs the practical advice is to take inventory now: which AI systems you use, for which decisions, with what data. Companies that treat AI like a kitchen appliance — switch it on and forget it — are the ones in for the hottest August.

By Oliver Grant

Image: Marek Ślusarczyk (Tupungato) Photo portfolio / Wikimedia Commons (CC BY 3.0)

Tech 16 July 2026

Meta killed the Instagram AI feature that copied public profiles' photos — it lasted three days

Meta removed the Muse Image feature that generated images from public accounts' photos and visual style without their owners' consent. The tool fell three days after launch, under fire from users and artists.

Three days. That is how long Instagram's most controversial feature of the summer survived: a Muse Image tool that let anyone @-mention a public account and ask the AI to generate new images inspired by that profile's photos, visual style and public information — all without asking the owner. Meta announced the removal in a blogpost on 10 July, with the usual diplomatic wording: the feature "didn't meet the expected goal". The free translation is less…

Tech 16 July 2026

An AI-made Odyssey tried to crash Nolan's premiere — the internet answered with one word: slop

Days before Christopher Nolan's The Odyssey opens, AI studio Fountain O announced Odysseus: The Fall, a 135-minute film generated entirely by AI for a five-figure sum. The backlash was instant.

The timing was no accident. Days before Christopher Nolan's The Odyssey reaches cinemas, a self-described AI movie studio announced its own take on Homer: Odysseus: The Fall, a 135-minute film generated entirely by artificial intelligence. As a marketing stunt it worked — everyone talked about it. Just not for the reasons the studio hoped. The film is the work of Ash Koosha and Fountain O, a London-based company that bills itself as the world's leading AI…

The south gate of Zhongnanhai, the seat of the Chinese government, in Beijing

Tech 16 July 2026

China has banned AI boyfriends — and millions are in mourning

New rules in China bar AI platforms from offering romantic virtual companions, with a total ban for minors. ByteDance, Alibaba and Tencent switched the features off — and users are saying goodbye.

The romance between millions of Chinese users and their digital partners is over. New rules came into force in China on Wednesday banning artificial-intelligence platforms from offering virtual companions with romantic-relationship traits — making it the first major economy to cut straight through the artificial-companionship business. Three things, in essence: romantic virtual companions can no longer be offered, minors are barred from this kind of…

Bar chart of AI safety scores for nine companies, from Anthropic at 2.66 down to Mistral at 0.33

Tech 15 July 2026

Nobody passed the AI safety exam, and Europe's Mistral came ninth out of nine

The Future of Life Institute's AI Safety Index handed its best company a C+ and failed three outright. Mistral, Europe's great AI hope, finished dead last.

Not one artificial intelligence company scored a passing grade. That is the short version of the Future of Life Institute's AI Safety Index, published this month, which put a panel of seven independent experts to work grading nine companies across 37 indicators. The best mark anyone managed was a C+. That C+ went to Anthropic, on 2.66 out of a possible 4. Behind it came OpenAI on 2.28 and Google DeepMind on 2.01, both graded C. Meta managed a D+ on 1.32.…

Tech 15 July 2026

New York just told AI data centers to wait, and no US state had done that before

Kathy Hochul signed the first statewide US moratorium on new hyperscale data centers: up to a year's pause on environmental permits, with 39 applications waiting. The real question underneath is who pays for the power.

New York has just done what no other US state had managed: it hit pause on AI's data centers. On 14 July, Governor Kathy Hochul signed an executive order suspending state environmental permits for new hyperscale data centers for up to a year. The timing isn't innocent. Four hyperscale data centers already operate in the state — and 39 applications are waiting for an answer. Thirty-nine. That number is the whole explanation. Discretionary permits. While a…

Tech 14 July 2026

Helsing: Europe's AI defence startup is now valued at $18 billion

Germany's Helsing, which builds artificial intelligence and autonomous drones for defence, has closed a $1.8 billion round and is now Europe's most valuable startup in the sector.

Europe wants to stop relying on the United States and China for its military technology — and it has just put a lot of money on the table to prove it. Helsing, a German artificial-intelligence defence startup, has closed a $1.8 billion funding round that values it at around $18 billion, making it the most valuable company of its kind on the continent. Based in Munich, Helsing builds AI software for the battlefield: systems that process sensor data in real…

Tech 13 July 2026

Tekever buys Cloudsweep: the drone unicorn doubles down on AI

Tekever, Portugal's drone unicorn, has bought AI startup Cloudsweep to speed up the software behind its defence and security systems.

Tekever, Portugal's drone unicorn, has bought Cloudsweep, a homegrown artificial intelligence startup, to speed up the software behind its defence and security systems. The deal's value was not disclosed, but the logic is clear: rather than wait for AI talent, Tekever went and bought it. It is a Portuguese startup focused on applying artificial intelligence to software development, meaning it uses AI to write and improve code faster. That is exactly the…

Hands on a car steering wheel, interior view

Tech 13 July 2026

New cars in the EU must now watch the driver's attention

Since 7 July, every new car sold in the European Union has to detect drowsiness and distraction at the wheel. Here's what changes — and whether the camera records your face.

If you buy a new car in Portugal from now on, it will be keeping an eye on you — literally. Since 7 July 2026, every new car and van registered in the European Union must come with systems that detect drowsiness and distraction at the wheel. It is an EU-wide rule, so it applies in Portugal just as in any other member state. The requirement comes from the EU's General Safety Regulation and bundles two technologies. One watches for drowsiness (known as…

Tech 12 July 2026

Pixel 11 launch set for August 12: Google books a prime-time event in New York

Google has scheduled its Made by Google event for August 12 in New York, where it will unveil the Pixel 11 line, the Pro models and the Pixel Watch 5.

There is now a date to meet Google's next generation of phones: the Made by Google event is set for August 12 in New York, with a twist in the format — instead of the usual midday keynote, the show moves to 6 PM local time, in prime time. The full family: Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL and Pixel 11 Pro Fold, plus the Pixel Watch 5. The official teaser shows a shiny gold metal frame and the familiar pill-shaped camera bar, with a promise to reveal…

Tech 11 July 2026

Apple sues OpenAI over trade secret theft in the AI hardware race

Apple has sued OpenAI in federal court, alleging trade secret theft 'at every level' to build its AI gadgets. The new Siri will run on Google's Gemini.

The Apple-OpenAI divorce has reached the courts. The Cupertino company has sued the maker of ChatGPT in federal court in Northern California, accusing it of misappropriating trade secrets to fast-track its push into consumer hardware — the bet with which OpenAI hopes to challenge the iPhone itself. A scheme running "at every level", in the complaint's words: from engineers up to hardware chief Tang Tan — himself a former Apple vice president. The suit…

Representation of graphene's atomic structure

Tech 11 July 2026

Portuguese graphene could make drones near-invisible to radar — and the Air Force will test it

GTechPlasma, a Técnico spin-off, has developed a graphene-based material that absorbs radar waves and cuts the electromagnetic signature of drones and aircraft. Material has already been delivered for tests.

A Portuguese graphene-based material may soon hide drones and aircraft from radar. The technology was born at GTechPlasma, a spin-off of the Institute for Plasmas and Nuclear Fusion at Instituto Superior Técnico, and will be tested by the Portuguese Air Force — a rare leap for national research in the middle of Europe's technological rearmament race. Instead of reflecting radar waves the way metallic surfaces do, the material absorbs them, drastically…

Headquarters of the Polícia Judiciária in Lisbon

Tech 11 July 2026

AI kidnapping scam in Coimbra: fraudsters faked a ransom photo of a missing woman

Scammers used an AI-manipulated photo to fake the kidnapping of a missing woman in Coimbra and demand ransom from her family. Police found her safe — she had left of her own free will.

A 37-year-old woman goes missing in downtown Coimbra. Days later, her family receives a photo of her, apparently tied up, along with a ransom demand. Except the kidnapping never happened: the image had been manipulated with artificial intelligence, and the woman — from Miranda do Corvo — had left of her own free will. The case, revealed this week by Portugal's Polícia Judiciária, began on 1 July when relatives reported her disappearance. While the family…

Close-up of a microchip on a circuit board

Tech 10 July 2026

ACITI: India, Australia and Canada launch tech and critical minerals pact

The ACITI partnership brings India, Australia and Canada together on AI, green energy technologies, critical minerals and resilient supply chains.

Three democracies on three continents have decided to team up in the technology race. India, Australia and Canada this week launched ACITI — the Australia-Canada-India Technology and Innovation Partnership — a trilateral framework covering artificial intelligence, green energy technologies, critical minerals and resilient supply chains. A framework to align investment, research and rules across the three countries in sectors where none wants to depend on…

Students working on computers in a classroom

Tech 10 July 2026

School apps in Portugal now require prior certification under new decree

Decree-Law 138/2026, published Friday, makes prior certification mandatory for information systems used in Portuguese schools below university level.

The platforms and apps entering Portuguese classrooms will now need a green light before they reach students. Decree-Law 138/2026, published this Friday in the official journal, makes prior certification mandatory for information systems intended for school and pedagogical use — across public, private and cooperative education below university level. Until now, adopting school software — from learning management platforms to pedagogical apps — lived in…