Breaking AI & Tech News Analyzed

Social media users in the UK are posting, sharing, and commenting less on tech platforms, driven by the rise of video-oriented apps and fears that online posts could harm their reputation. According to Ofcom, 49% of adult social media users now post, share, or comment, down from 61% in 2024.The proportion of users exploring new websites has also decreased, from 70% to 56%. Joseph Oxlade, senior research manager at Ofcom, cited the popularity of video apps like TikTok and Instagram Reels as a reason for the decline in active use.Concerns about the long-term impact of online posts on personal and professional lives are also a factor. 49% of adults are now concerned about posts causing them problems in the future, up from 43% in 2024. This fear is not unfounded, as historic internet posts have been known to cause embarrassment for public figures.Despite this, social media use remains widespread, with 89% of adult internet users using at least one social media platform. The Ofcom data was based on a survey of 7,500 people across the UK last year over the age of 16.The data also showed that use of AI tools like ChatGPT has increased, with 54% of UK adults using them, up from 31% in 2024. Some users are interacting with AI as if it were a person, using it for tasks like seeking relationship advice or generating creative content.

The Lead Apple has rolled out critical security updates for older iPhone and iPad models to counter a sophisticated web-based attack known as DarkSword. The release of iOS 18.7.7 and iPadOS 18.7.7 is a direct response to a leaked set of hacking tools that can compromise devices running versions 18.4 through 18.7. Understanding the DarkSword Vulnerability DarkSword is a sophisticated exploit kit that operates through a 'drive-by download' mechanism. Attackers do not need to trick users into clicking suspicious links; instead, simply visiting a legitimate website that has been breached can trigger the malicious code. This allows the toolkit to break into Apple devices and install spyware without the user's immediate knowledge. The Data Impact of the Exploit The capabilities of the DarkSword toolkit pose a significant threat to user privacy. Once a device is compromised, attackers gain access to a wide range of sensitive information, including: Private messages Browser history Location data Cryptocurrency wallet credentials Security researchers have observed these tools being used in targeted attacks across China, Malaysia, Turkey, Saudi Arabia, and Ukraine. User Friction and Update Resistance Despite the severity of the threat, Apple notes that millions of users remain vulnerable because they have chosen not to update their devices. The primary driver for this resistance is the user experience; many users have opted out of the latest software updates to avoid the new 'liquid glass' interface, prioritizing familiarity over security patches. The Role of Lockdown Mode For users who remain at high risk, Apple’s optional Lockdown Mode offers a robust defense. The company has confirmed that this feature effectively blocks attacks that would bypass standard protections, including those from government-sponsored spyware campaigns. Future Outlook on Web-Based Threats The publication of the DarkSword toolkit on the open web signals a worrying trend. As these tools become more accessible, we can expect an increase in low-cost, high-impact cyberattacks targeting older device versions that lack the latest security protocols.

The author spent a month testing Meta's smartglasses, which feature an integrated AI assistant voiced by Judi Dench. The glasses can take photos, provide directions, and answer questions, but the author found them to be unreliable and frustrating to use.The glasses raise significant privacy concerns, with the author noting that they can be used for covert recording and that people may not be aware when they are being filmed. The author also experienced creepy feelings while wearing the glasses, particularly when they were used to record strangers without their consent.The author concludes that while smartglasses have the potential to be useful assistive technology for people with disabilities, they are not yet reliable or functional enough to be widely adopted. The author also notes that Meta's plans for facial recognition and data collection raise significant concerns about privacy and surveillance.Experts warn that smartglasses may become a flashpoint in a bigger, existential discussion about how much integration we want with technology, and that regulation is needed to protect bystanders' privacy. The author ultimately decides that the glasses are not worth the risks and drawbacks, and that people should think carefully before adopting this technology.

Police in Wuhan reported a sudden "system malfunction" that immobilised several autonomous robotaxis operated by Baidu’s Apollo Go service, leaving passengers stuck on an elevated highway for up to an hour and a half.Local authorities said they received a flood of calls on Tuesday night from riders whose vehicles froze in the middle of the road. A police statement confirmed that “multiple Apollo Go cars stopped in the middle of the road, unable to move,” and preliminary investigations point to a technical failure.Baidu maintains a fleet of more than 500 driverless cars in Wuhan, though the exact number affected was not disclosed. One commuter shared a 90‑minute ordeal on the Chinese platform RedNote, describing how the vehicle stalled at 9 p.m. on an overpass, surrounded by dump trucks, while customer‑service lines remained unanswered.The rider eventually was rescued after the order was cancelled at 10:30 p.m., but criticized Apollo Go’s support team for offering “useless platitudes” instead of concrete solutions. Social‑media users also posted videos captioned “Apollo Go, are you paralysed?” showing futile attempts to contact the company via the in‑car tablet.This is not Baidu’s first controversy. In December, authorities in Zhuzhou halted robotaxi operations after a Baidu‑manufactured autonomous vehicle struck two pedestrians, sending them to intensive care.Despite these setbacks, Baidu’s autonomous‑mobility arm continues to grow. Company filings reveal that Apollo Go delivered 3.4 million driverless rides in the fourth quarter of 2025, a jump of over 200 % compared with the same period in 2024. The firm is also pursuing international expansion, having announced partnership deals with rideshare giants Lyft and Uber to deploy its vehicles on their platforms.When approached for comment, Baidu did not respond, according to Reuters.Additional reporting by Yu‑chen Li

Apple has launched its latest laptop, the MacBook Neo, which marks a significant departure from its traditional Mac lineup. The Neo is powered by an A18 Pro chip typically found in iPhones, offering a highly performant experience, particularly in everyday tasks.The MacBook Neo starts at £599 (€699/$599/A$899), making it an attractive option for those looking for a budget-friendly laptop without sacrificing performance. The laptop features a 13in LCD screen with a resolution of 2408x1506 and 219 ppi, 8GB of RAM, and storage options of 256GB or 512GB.Despite its lower price point, the Neo doesn't feel like a cut-price machine. The laptop has a high-quality aluminium body, sleek lines, and rounded corners, similar to the MacBook Air and Pro models. The keyboard and trackpad are also top-notch, with the trackpad being mechanical rather than haptic.The Neo's performance is impressive, handling complex edits to multiple large file photos in Affinity and Pixelmator Pro without breaking a sweat. It's also capable of light video editing, making it suitable for everyday tasks.The laptop's battery life is excellent, managing about 13 hours between charges for general browsing, note-taking, image editing, chat, and emails. The Neo also features a 1080p webcam, good mics, and decent speakers.However, there are some limitations compared to the more expensive MacBook Air, such as no wifi 7, no Thunderbolt, no 5K display support, and only 8GB of memory. Nevertheless, the MacBook Neo sets a new benchmark for budget-friendly laptops, making it an attractive option for students and those on a budget.

The Erosion of Digital Anonymity Apple's 'Hide My Email' feature, designed to shield user identities from apps and websites, has been exposed as ineffective against federal subpoenas. The company recently revealed it provided real names and email addresses to the FBI and ICE, undermining the feature's promise of anonymity for paying iCloud+ subscribers. This disclosure highlights a critical vulnerability in the privacy architecture of major tech platforms, where 'anonymity' often depends on the willingness of the provider to withhold data. The 'Hide My Email' Loophole The feature allows iCloud+ subscribers to generate anonymous email aliases that forward messages to their private inbox. While Apple claims it does not read the content of these forwarded messages, the legal mechanism allows authorities to bypass the alias entirely. In a recent affidavit, the FBI revealed that Apple provided the real identity behind an anonymized address used in a threat investigation against Kash Patel's girlfriend. Similarly, ICE agents obtained records linking multiple anonymized accounts to a specific individual involved in an alleged identity fraud scheme. Metadata vs. Content The data shared with law enforcement goes beyond simple forwarding logs; Apple provided the account holder's full name, email address, and billing information. In one instance, Apple disclosed records for 134 anonymized email accounts created via the feature. This indicates that while the content of emails remains private, the ownership of the account is easily accessible to authorities with a valid legal request. The distinction between encrypted content and unencrypted metadata is becoming the primary battleground for digital privacy. End-to-End Encryption Limits This incident underscores a critical distinction in modern cybersecurity: the difference between end-to-end encryption (E2EE) and account metadata. Apple touts its services as E2EE, meaning only the user can access their data. However, this protection does not extend to the account registration details, billing history, and unencrypted routing information that Apple stores. As a result, the demand for alternative privacy tools like Signal, which offer stronger protections against metadata collection, is likely to increase among privacy-conscious users. The Future of Privacy vs. Security As law enforcement agencies increasingly rely on metadata to solve crimes, tech companies will face mounting pressure to balance user privacy with national security obligations. We can expect a rise in legal battles regarding the scope of 'anonymized' services and a potential shift in consumer behavior, where users seek out services that offer true anonymity rather than just obfuscation.

Ukraine's expertise in drone technology has garnered attention from countries in the Gulf region, which are looking to bolster their defense capabilities. The country's experience in using drones in conflict zones has made its technology and know-how an attractive proposition for Gulf nations. Ukraine's drone industry has seen significant growth in recent years, driven in part by the need to counter threats on its own borders. This growth has led to the development of a range of drones, from small reconnaissance units to larger, more heavily armed models. Gulf countries, including Saudi Arabia and the United Arab Emirates, have been investing heavily in their military capabilities in recent years, seeking to modernize their defense systems and address emerging threats. The interest in Ukraine's drone technology is part of this broader effort to enhance their military capabilities. The potential for partnerships between Ukraine and Gulf countries in the field of drone technology could have significant implications for regional security dynamics. As these countries look to develop their own drone capabilities, they may also be seeking to reduce their reliance on traditional defense suppliers.

Meta's attempt to shift the blame for a young user's mental health issues away from its platforms and onto her family and offline social problems has backfired in a landmark social media addiction trial. The company, parent of Facebook and Instagram, employed a aggressive defense strategy that included highlighting the plaintiff's complaints about her mother in her teenage text messages and personal writings.The jurors, however, were not convinced by Meta's arguments and decided 10-2 in favor of the plaintiffs, awarding $4.2m in damages from Meta and $1.8m from co-defendant YouTube. This verdict could set a precedent for thousands of similar trials already in the works against social media companies.Meta's defense strategy drew backlash from parental advocates, who argued that the company was attempting to shift the blame away from its own design choices and onto parents and users. "For the biggest tech executives, I want to say something: stop blaming the parents. It's on you," said Julianna Arnold, a co-founder of advocacy group Parents Rise!.The trial's outcome reflects a growing distrust of social media companies and their impact on society. A Pew Research Survey found that around 64% of US adults believe that social media platforms have a negative impact on the country, and around two-thirds of Americans have a negative view of Meta CEO Mark Zuckerberg.The verdict has been celebrated by advocates for reining in big tech, who see it as a significant step towards holding social media companies accountable for their role in shaping societal harms. "This trial was proof that if you put CEOs like Mark Zuckerberg on the stand before a judge and jury of their peers, the tech industry's wanton disregard for people will be on full display," said Sacha Haworth, executive director of The Tech Oversight Project.

The literary world is grappling with a new challenge: detecting AI-written books. Kate Nash, a literary agent, recently discovered an AI prompt at the top of a submission letter, which raised concerns about the use of artificial intelligence in book writing. The controversy surrounding Mia Ballard's 'femgore' horror novel Shy Girl, which was suspected to be up to 78% AI-generated, has forced literary agents and publishers to reevaluate their ability to detect AI-generated work. Hachette halted US publication of Shy Girl and removed it from online retailers after an internal review. Experts warn that AI detection tools are not foolproof and that the technology is advancing rapidly. Prof Patrick Juola, a US computer scientist, likens the failure of AI detection tools to antibiotic resistance, stating that AI is a learning system continually upgraded by its manufacturers. Mor Naaman, a professor of information science at Cornell Tech, agrees that publishers won't stand a chance in detecting AI-generated books soon. He notes that AI learns quickly how to avoid detection and that the line between human and AI-generated content is becoming increasingly blurred. The debate raises questions about the role of AI in creative writing and the importance of trust between readers and writers. Anna Ganley, of the Society of Authors, launched the Human Authored scheme to identify works written by humans, but it relies on trust, a value that is inherently vulnerable. Amelia Hill, the author of the article, highlights the importance of preserving the human element in literature, stating that AI may flood the market with content, but it cannot replace the messy, difficult work of being human.