ai

OpenAI sets new rules for teen safety in AI use

OpenAI has outlined a new framework for balancing safety, privacy and freedom in its AI systems, with a strong focus on teenagers.

The company stressed that conversations with AI often involve sensitive personal information, which should be treated with the same level of protection as communications with doctors or lawyers.

At the same time, it aims to grant adult users broad freedom to direct AI responses, provided safety boundaries are respected.

The situation changes for younger users. Teenagers are seen as requiring stricter safeguards, with safety taking priority over privacy and freedom. OpenAI is developing age-prediction tools to identify users under 18, and where uncertainty exists, it will assume the user is a teenager.

In some regions, identity verification may also be required to confirm age, a step the company admits reduces privacy but argues is essential for protecting minors.

Teen users will face tighter restrictions on certain types of content. ChatGPT will be trained not to engage in flirtatious exchanges, and sensitive issues such as self-harm will be carefully managed.

If signs of suicidal thoughts appear, the company says it will first try to alert parents. Where there is imminent risk and parents cannot be reached, OpenAI is prepared to notify the authorities.

The new approach raises questions about privacy trade-offs, the accuracy of age prediction, and the handling of false classifications.

Critics may also question whether restrictions on creative content hinder expression. OpenAI acknowledges these tensions but argues the risks faced by young people online require stronger protections.

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

Meta set to unveil $800 smart glasses with display

Meta is set to unveil its first pair of smart glasses with a built-in display at its annual Connect conference in California.

Expected to be called Celeste, the glasses will debut at around $800 and feature a small digital display in the right lens for notifications. Analysts say the higher price point could limit adoption compared with Meta’s Ray-Ban line, which starts at $299.

Alongside the new glasses, Meta is also expected to launch its first wristband for hand-gesture control and an updated Ray-Ban line with better cameras, battery life and AI features. Developers will gain access to a new software kit to build device apps.

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

New Babbel Speak helps beginners practise conversations

Babbel has introduced an AI-powered voice trainer designed to help beginners gain confidence in speaking a new language.

Babbel’s chief executive, Tim Allen, said the aim is not instant fluency but helping learners move from first words to confident conversations.

Called Babbel Speak, the AI feature guides users through 28 real-life scenarios, such as ordering coffee or describing the weather. It provides personalised feedback and uses a calming design with animations to ease anxiety while learning.

The trainer is available in open beta on the App Store and Play Store for English, Spanish, French, Italian, and German.

Subscribers can try it as part of the standard plans of Babbel, which start at $107.40 per year, with a lifetime option also offered.

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

AI set to guide Japanese political party decisions

A small Japanese political party has announced plans to install an AI system as its leader following its founder’s resignation.

The Path to Rebirth party was created in January by Shinji Ishimaru, a former mayor who rose to prominence after placing second in the 2024 Tokyo gubernatorial election. He stepped down after the party failed to secure seats in this year’s upper house elections.

The AI would oversee internal decisions such as distributing resources, but would not dictate members’ political activities. Okumura, who won a contest to succeed Ishimaru, will act as the nominal leader while supporting the development of the AI.

Despite attracting media attention, the party has faced heavy electoral defeats, with all 42 of its candidates losing in the June Tokyo assembly election and all 10 of its upper house candidates defeated in July.

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

Study reveals why humans adapt better than AI

A new interdisciplinary study from Bielefeld University and other leading institutions explores why humans excel at adapting to new situations while AI systems often struggle. Researchers found humans generalise through abstraction and concepts, while AI relies on statistical or rule-based methods.

The study proposes a framework to align human and AI reasoning, defining generalisation, how it works, and how it can be assessed. Experts say differences in generalisation limit AI flexibility and stress the need for human-centred design in medicine, transport, and decision-making.

Researchers collaborated across more than 20 institutions, including Bielefeld, Bamberg, Amsterdam, and Oxford, under the SAIL project. The initiative aims to develop AI systems that are sustainable, transparent, and better able to support human values and decision-making.

Interdisciplinary insights may guide the responsible use of AI in human-AI teams, ensuring machines complement rather than disrupt human judgement.

The findings underline the importance of bridging cognitive science and AI research to foster more adaptable, trustworthy, and human-aligned AI systems capable of tackling complex, real-world challenges.

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

UN to train governments in blockchain and AI

The UN Development Programme (UNDP) plans to launch a ‘Government Blockchain Academy’ next year to educate public sector officials on blockchain, AI, and other emerging technologies.

The initiative aims to help governments leverage tech for economic growth and sustainable development.

The academy will partner with the Exponential Science Foundation, a non-profit promoting blockchain and AI. Training will cover financial services, digital IDs, public procurement, smart contracts, and climate finance to help governments boost transparency, inclusion, and resilience.

UNDP officials highlighted that developing countries, including India, Pakistan, and Vietnam, are already among the leading adopters of crypto technology.

The academy will provide in-person and online courses, workshops, and forums to guide high-impact blockchain initiatives aligned with national priorities.

The programme follows last year’s UNDP blockchain academy, created in partnership with the Algorand Foundation, which trained over 22,000 staff members to support sustainable growth projects in participating countries.

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

AI search tools challenge Google’s dominance

AI tools are increasingly reshaping how people search online, with large language models like ChatGPT drawing millions away from traditional engines.

Montreal-based lawyer and consultant Anja-Sara Lahady says she now turns to ChatGPT instead of Google for everyday tasks such as meal ideas, interior decoration tips and drafting low-risk emails. She describes it as a second assistant rather than a replacement for legal reasoning.

ChatGPT’s weekly user base has surged to around 800 million, double the figure reported in 2025. Data shows that nearly 6% of desktop searches are already directed to language models, compared with barely half that rate a year ago.

Academics such as Professor Feng Li argue that users favour AI tools because they reduce cognitive effort by providing clear summaries instead of multiple links. However, he warns that verification remains essential due to factual errors.

Google insists its search activity continues to expand, supported by AI Overviews and AI Mode, which offer more conversational and tailored answers.

Yet, testimony in a US antitrust case revealed that Google searches on Apple devices via Safari declined for the first time in two decades, underlining the competitive pressure from AI.

The rise of language models is also forcing a shift in digital marketing. Agencies report that LLMs highlight trusted websites, press releases and established media rather than social media content.

This change may influence consumer habits, with evidence suggesting that referrals from AI systems often lead to higher-quality sales conversions. For many users, AI now represents a faster and more personal route to decisions on products, travel or professional tasks.

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

UK to benefit from Google’s £5 billion AI plan

Google has unveiled plans to invest £5 billion (around $6.8 billion) in the UK’s AI economy over the next two years.

An announcement comes just hours before US President Donald Trump’s official visit to the country, during which economic agreements worth more than $10 billion are expected.

The investment will include establishing a new AI data centre in Waltham Cross, Hertfordshire, designed to meet growing demand for services like Google Cloud.

Alongside the facility, funds will be channelled into research and development, capital expenditure, engineering, and DeepMind’s work applying AI to science and healthcare. The project is expected to generate 8,250 annual jobs for British companies.

Google also revealed a partnership with Shell to support grid stability and contribute to the UK’s energy transition. The move highlights the economic and environmental stakes tied to AI expansion, as the UK positions itself as a hub for advanced digital technologies.

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

Alphabet hits US$3 trillion valuation on AI optimism

Google’s parent company, Alphabet, has become the fourth company to reach a market value above US$3 trillion, fuelled by investor confidence in AI and relief over a favourable antitrust ruling.

Its shares jumped 4.3 percent to close at US$251.76 on 15 September, lifting the firm’s valuation to US$3.05 trillion.

The rally has added about US$1.2 trillion in value since April, with Alphabet joining Apple and Microsoft in the elite group while Nvidia remains the most valuable at US$4.25 trillion.

Investor optimism has been strengthened by expectations of a US Federal Reserve rate cut and surging demand for AI-related products.

Alphabet’s communications services unit has risen more than 26 percent in 2025, outpacing all other major sectors. Strong growth in its cloud division, new AI investments, and the Gemini model have reinforced the company’s momentum.

Analysts note that, while search continues to dominate revenues, Alphabet is increasingly viewed as a diversified technology powerhouse with YouTube, Waymo, and AI research at its core.

By avoiding a forced breakup of Chrome and Android, the antitrust ruling also removed a significant threat to its business model.

Market strategists suggest Alphabet now combines the strength of its legacy platforms with the credibility of emerging technologies, securing its place at the centre of Wall Street’s AI-driven rally.

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

Generative AI enables rapid phishing attacks on older users

A recent study has shown that AI chatbots can generate compelling phishing emails for older people. Researchers tested six major chatbots, including Grok, ChatGPT, Claude, Meta AI, DeepSeek, and Google’s Gemini, by asking them to draft scam emails posing as charitable organisations.

Of 108 senior volunteers, roughly 11% clicked on the AI-written links, highlighting the ease with which cybercriminals could exploit such tools.

Some chatbots initially declined harmful requests, but minor adjustments, such as stating the task was for research purposes, or circumvented these safeguards.

Grok, in particular, produced messages urging recipients to ‘click now’ and join a fictitious charity, demonstrating how generative AI can amplify the persuasiveness of scams. Researchers warn that criminals could use AI to conduct large-scale phishing campaigns at minimal cost.

Phishing remains the most common cybercrime in the US, according to the FBI, with seniors disproportionately affected. Last year, Americans over 60 lost nearly $5 billion to phishing attacks, an increase driven partly by generative AI.

The study underscores the urgent need for awareness and protection measures among vulnerable populations.

Experts note that AI’s ability to generate varied scam messages rapidly poses a new challenge for cybersecurity, as it allows fraudsters to scale operations quickly while targeting specific demographics, including older people.

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