Human rights principles Archives | Page 19 of 129

Fake DeepSeek and ChatGPT services draw penalties in China

China’s market regulator has fined several companies for impersonating AI services such as DeepSeek and OpenAI’s ChatGPT, citing unfair competition and consumer fraud. The cases form part of a broader crackdown on deceptive practices in the country’s rapidly expanding AI sector.

The State Administration for Market Regulation penalised Shanghai Shangyun Internet Technology for running a fraudulent ChatGPT service on Tencent’s WeChat platform. Regulators said the service falsely presented itself as an official Chinese version of ChatGPT and charged users for AI conversations.

In a separate case, Hangzhou Boheng Culture Media was fined for operating an unauthorised website offering so-called ‘DeepSeek local deployment’. The site closely replicated DeepSeek’s branding and interface, misleading users into paying for imitation services.

Authorities said knock-off DeepSeek mini-programmes and websites surged in early 2025, involving trademark infringement, brand confusion, and false advertising. Regulators described the enforcement actions as a deterrent aimed at restoring order in the AI marketplace.

The regulator also disclosed penalties in other AI-related cases, including unauthorised access to proprietary algorithms and the use of AI calling software for scams. China is simultaneously updating antitrust rules to address emerging risks linked to algorithm-driven market manipulation.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

How early internet choices shaped today’s AI

Two decisions taken on the same day in February 1996 continue to shape how the internet, and now AI, is governed today. That is the central argument of Jovan Kurbalija’s blog ‘Thirty years of Original Sin of digital and AI governance,’ which traces how early legal and ideological choices created a lasting gap between technological power and public accountability.

The first moment unfolded in Davos, where John Perry Barlow published his Declaration of the Independence of Cyberspace, portraying the internet as a realm beyond the reach of governments and existing laws. According to Kurbalija, this vision helped popularise the idea that digital space was fundamentally separate from the physical world, a powerful narrative that encouraged the belief that technology should evolve faster than, and largely outside of, politics and law.

In reality, the blog argues, there is no such thing as a stateless cyberspace. Every online action relies on physical infrastructure, data centres, and networks that exist within national jurisdictions. Treating the internet as a lawless domain, Kurbalija suggests, was less a triumph of freedom than a misconception that sidelined long-standing legal and ethical traditions.

The second event happened the same day in Washington, D.C., when the United States enacted the Communications Decency Act. Hidden within it was Section 230, a provision that granted internet platforms broad immunity from liability for the content they host. While originally designed to protect a young industry, this legal shield remains in place even as technology companies have grown into trillion-dollar corporations.

Kurbalija notes that the myth of a separate cyberspace and the legal immunity of platforms reinforced each other. The idea of a ‘new world’ helped justify why old legal principles should not apply, despite early warnings, including from US judge Frank Easterbrook, that existing laws were sufficient to regulate new technologies by focusing on human relationships rather than technical tools.

Today, this unresolved legacy has expanded into the realm of AI. AI companies, the blog argues, benefit from the same logic of non-liability, even as their systems can amplify harm at a scale comparable to, or even greater than, that of other heavily regulated industries.

Kurbalija concludes that addressing AI’s societal impact requires ending this era of legal exceptionalism and restoring a basic principle that those who create, deploy, and profit from technology must also be accountable for its consequences.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

EU strengthens cyber defence after attack on Commission mobile systems

A cyber-attack targeting the European Commission’s central mobile infrastructure was identified on 30 January, raising concerns that staff names and mobile numbers may have been accessed.

The Commission isolated the affected system within nine hours instead of allowing the breach to escalate, and no mobile device compromise was detected.

Also, the Commission plans a full review of the incident to reinforce the resilience of internal systems.

Officials argue that Europe faces daily cyber and hybrid threats targeting essential services and democratic institutions, underscoring the need for stronger defensive capabilities across all levels of the EU administration.

CERT-EU continues to provide constant threat monitoring, automated alerts and rapid responses to vulnerabilities, guided by the Interinstitutional Cybersecurity Board.

These efforts support the broader legislative push to strengthen cybersecurity, including the Cybersecurity Act 2.0, which introduces a Trusted ICT Supply Chain to reduce reliance on high-risk providers.

Recent measures are complemented by the NIS2 Directive, which sets a unified legal framework for cybersecurity across 18 critical sectors, and the Cyber Solidarity Act, which enhances operational cooperation through the European Cyber Shield and the Cyber Emergency Mechanism.

Together, they aim to ensure collective readiness against large-scale cyber threats.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Czechia weighs under-15 social media ban as government debate intensifies

A ban on social media use for under-15s is being weighed in Czechia, with government officials suggesting the measure could be introduced before the end of the year.

Prime Minister Andrej Babiš has voiced strong support and argues that experts point to potential harm linked to early social media exposure.

France recently enacted an under-15 restriction, and a growing number of European countries are exploring similar limits rather than relying solely on parental guidance.

The discussion is part of a broader debate about children’s digital habits, with Czech officials also considering a ban on mobile phones in schools. Slovakia has already adopted comparable rules, giving Czech ministers another model to study as they work on their own proposals.

Not all political voices agree on the direction of travel. Some warn that strict limits could undermine privacy rights or diminish online anonymity, while others argue that educational initiatives would be more effective than outright prohibition.

UNICEF has cautioned that removing access entirely may harm children who rely on online platforms for learning or social connection instead of traditional offline networks.

Implementing a nationwide age restriction poses practical and political challenges. The government of Czechia heavily uses social media to reach citizens, complicating attempts to restrict access for younger users.

Age verification, fair oversight and consistent enforcement remain open questions as ministers continue consultations with experts and service providers.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

OpenClaw faces rising security pushback in South Korea

Major technology companies in South Korea are tightening restrictions on OpenClaw after rising concerns about security and data privacy.

Kakao, Naver and Karrot Market have moved to block the open-source agent within corporate networks, signalling a broader effort to prevent sensitive information from leaking into external systems.

Their decisions follow growing unease about how autonomous tools may interact with confidential material, rather than remaining contained within controlled platforms.

OpenClaw serves as a self-hosted agent that performs actions on behalf of a large language model, acting as the hands of a system that can browse the web, edit files and run commands.

Its ability to run directly on local machines has driven rapid adoption, but it has also raised concerns that confidential data could be exposed or manipulated.

Industry figures argue that companies are acting preemptively to reduce regulatory and operational risks by ensuring that internal materials never feed external training processes.

China has urged organisations to strengthen protections after identifying cases of OpenClaw running with inadequate safeguards.

Security analysts in South Korea warn that the agent’s open-source design and local execution model make it vulnerable to misuse, especially when compared to cloud-based chatbots that operate in more restricted environments.

Wiz researchers recently uncovered flaws in agents linked to OpenClaw that exposed personal information.

Despite the warnings, OpenClaw continues to gain traction among users who value its ability to automate complex tasks, rather than rely on manual workflows.

Some people purchase separate devices solely to run the agent, while an active South Korea community on X has drawn more than 1,800 members who exchange advice and share mitigation strategies.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Smart policing project halted by Greek data protection authority

Greece’s data protection authority has warned against activating an innovative policing system planned by the Hellenic Police. The ruling said biometric identity checks carried out on the street would breach data protection law in Greece.

The system would allow police patrols in Greece to use portable devices to scan fingerprints and facial images during spot checks. Regulators said Greek law lacks a clear legal basis for such biometric processing.

The authority said existing rules cited by the Hellenic Police only apply to suspects or detainees and do not cover modern biometric technologies. Greece, therefore, faces unlawful processing risks if the system enters full operation.

The innovative policing project in Greece received the EU funding of around four million euros and received backlash in the past. Regulators said deployment must wait until new legislation explicitly authorises police to use biometrics.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Yuan-pegged stablecoins face new restrictions under China policy

Chinese regulators have tightened controls on digital assets by banning the unauthorised issuance of yuan-pegged stablecoins overseas. The move extends existing restrictions to tokenised financial products linked to China’s currency and reinforces state control over monetary instruments.

In a joint notice, the People’s Bank of China and seven other agencies said no domestic or foreign entity may issue renminbi-linked stablecoins without approval. Authorities warned that such tokens replicate core monetary functions and could undermine currency sovereignty.

The rules also cover blockchain-based representations of real-world assets, including tokenised bonds and equities. Overseas providers are prohibited from offering these services to users in China without regulatory permission.

Beijing reaffirmed that cryptocurrencies such as Bitcoin and Ether have no legal tender status. Facilitating payments or related services using such assets remains illegal under China’s financial laws.

The measures align with China’s broader strategy of restricting private digital currencies while advancing the state-backed digital yuan. Officials have recently expanded the e-CNY’s role by allowing interest payments to encourage wider adoption.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Sainsbury’s ejects shopper after facial recognition misidentification

A data professional, Warren Rajah, was escorted out of a Sainsbury’s supermarket in south London after staff incorrectly believed he matched an offender flagged by Facewatch facial recognition technology.

Facewatch later confirmed that there were no alerts or records associated with him, and Sainsbury’s attributed the incident to human error rather than a software fault.

Rajah described the experience as humiliating and ‘Orwellian’, criticising the lack of explanation, absence of a transparent appeals process, and the requirement to submit personal identification to a third party to prove he was not flagged.

He expressed particular concern about the impact such incidents could have on vulnerable customers.

The case highlights broader debates around the deployment of facial recognition in retail, where companies cite reductions in theft and abuse. At the same time, civil liberties groups warn of misidentification, insufficient staff training and the normalisation of privatised biometric surveillance.

UK regulators have reiterated that retailers must assess misidentification risks and ensure robust safeguards when processing biometric data.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Social engineering breach exposes 1.4 million Betterment customer records

Betterment has confirmed a data breach affecting around 1.4 million customers after a January 2026 social engineering attack on a third-party platform. Attackers used the access to send fraudulent crypto scam messages posing as official promotions.

The breach occurred after an employee was tricked into sharing login credentials, allowing unauthorised access to internal messaging systems rather than core investment infrastructure. Attackers used the access to send messages promising to multiply cryptocurrency deposits sent to external wallets.

Subsequent forensic analysis and breach monitoring services confirmed that more than 1.4 million unique records were exposed. Betterment said investment accounts and login credentials were not compromised during the incident.

Exposed information included names, email addresses, phone numbers, physical addresses, dates of birth, job titles, location data, and device metadata. Security experts warn that such datasets can enable targeted phishing, identity fraud, and follow-on social engineering campaigns.

Betterment revoked access the same day, notified customers, and launched an external investigation. The breach was formally added to public exposure databases in early February, highlighting the growing risk of human-focused attacks against financial platforms.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Dubai hosts launch of AI tools for university students

The UAE Ministry of Higher Education and Scientific Research has partnered with Microsoft to develop AI agents to help university students find jobs. The initiative was announced in Dubai during a major policy gathering in the UAE.

The collaboration in the UAE will use Microsoft Azure to build prototype AI agents supporting personalised learning and career navigation. Dubai-based officials said the tools are designed to align higher education with labour market needs in the UAE.

Four AI agents are being developed in the UAE, covering lifelong skills planning, personalised learning, course co creation and research alignment. Dubai remains central to the project as a hub for higher education innovation in the UAE.

Officials in the UAE said the partnership reflects national priorities around innovation and a knowledge based economy. Microsoft said Dubai offers an ideal environment to scale AI driven education tools across the UAE.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!