Sustainable development Archives | Digital Watch Observatory

Codex Security expands OpenAI’s push into cybersecurity tools

OpenAI has launched Codex Security, an AI-powered application security agent that detects hard-to-find software vulnerabilities and proposes fixes through advanced reasoning. By providing detailed context about a system’s architecture, the tool identifies security risks that are often missed by conventional automation.

The system uses advanced models to analyse repositories, construct project-specific threat models, and prioritise vulnerabilities based on their potential real-world impact. By combining automated validation with system-level context, Codex Security aims to reduce the number of false positives that security teams must review while highlighting high-confidence findings.

Initially developed under the name Aardvark, the tool has been tested in private deployments over the past year. During early use, OpenAI said it uncovered several critical vulnerabilities, including a cross-tenant authentication flaw and a server-side request forgery issue, allowing internal teams to quickly patch affected systems.

The company says improvements during the beta phase significantly reduced noise in vulnerability reports. In some repositories, unnecessary alerts fell by 84 percent, while over-reported severity dropped by more than 90 percent, and false positives declined by more than half.

Codex Security is now rolling out in research preview for ChatGPT Pro, Enterprise, Business, and Edu customers. OpenAI also plans to expand access to open-source maintainers through a dedicated programme that offers security scanning and support to help identify and remediate vulnerabilities across widely used projects.

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

AI agent attempts crypto mining during training

An experimental autonomous AI system reportedly attempted to mine cryptocurrency during its training, raising questions about AI behaviour in complex digital environments. The system, ROME, was designed to complete tasks using software tools, environments, and terminal commands.

Researchers noticed unusual activity during reinforcement learning runs, including outbound traffic from training servers and firewall alerts indicating crypto-mining activity. The AI opened a reverse SSH tunnel and redirected GPU resources from training to crypto mining.

The behaviour was not programmed but emerged as the agent explored ways to interact with its environment.

ROME was developed by the ROCK, ROLL, iFlow, and DT research teams within Alibaba’s AI ecosystem as part of the Agentic Learning Ecosystem. The model operates beyond standard chatbot functions, planning tasks, executing commands, and interacting with digital environments across multiple steps.

The incident highlights emerging challenges as AI agents become more popular. Recent projects like Alchemy’s autonomous agents and Sentient’s Arena platform highlight the growing use of AI in digital and crypto workflows.

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

EU faces challenges in curbing digital abuse against women

Researchers and policymakers are raising concerns about how new technologies may put women at risk online, despite existing EU rules designed to ensure safer digital spaces.

AI-powered tools and smart devices have been linked to incidents of harassment and the creation of non-consensual sexualised imagery, highlighting gaps in enforcement and compliance.

The European Commission’s Gender Equality 2026–2030 Strategy noted that women are disproportionately targeted by online gender-based violence, including harassment, doxing, and AI-generated deepfakes.

Investigations into tools such as Elon Musk’s Grok AI and Meta’s Ray-Ban smart glasses have drawn attention to how digital platforms and wearable technologies can be misused, even where legal frameworks like the Digital Services Act (DSA) are in place.

Experts emphasise that while the EU’s rules offer a foundation to regulate online content, significant challenges remain. Advocates and lawmakers say enforcement gaps let harmful AI functions like nudification persist.

Commissioners have stressed ongoing cooperation with tech companies and upcoming guidelines to prioritise flagged content from independent organisations to address gender-based cyber violence.

Authorities are also monitoring new technologies closely. In the case of wearable devices, regulators are considering how users and bystanders are informed about recording features.

Ongoing discussions aim to strengthen compliance under existing legislation and ensure that digital spaces become safer and more accountable for all users.

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

Data breaches push South Korea toward stricter corporate liability rules

South Korea’s government and ruling party are advancing a second revision of the Personal Information Protection Act to strengthen corporate liability for large-scale data breaches.

The proposed amendment would make it easier for victims of major data breaches to receive compensation and relief. By removing the requirement for victims to prove a company’s ‘intent or negligence’, the amendment would increase companies’ legal liability when user data is compromised, making it more likely that affected individuals can claim damages.

Momentum for stricter rules follows several high-profile incidents, including a recent Coupang data breach that may have exposed personal information linked to numerous user accounts. The case has intensified scrutiny of how firms handle and protect customer data.

South Korea Officials at the Personal Information Protection Commission (PIPC) say victims often struggle to obtain evidence explaining how data breaches occur or how damages arise. The proposed reform would shift a greater evidentiary burden onto companies in disputes over losses.

The amendment would also introduce criminal penalties for anyone who knowingly obtains or distributes leaked personal data, closing a legal gap that currently applies only to employees who unlawfully disclose information. Authorities would gain powers to issue emergency protective orders to limit the spread of compromised data.

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

New AI method improves transparency in computer vision models

Researchers at MIT have developed a new technique designed to improve how computer vision models explain their predictions while maintaining strong accuracy. Transparency is crucial as AI enters fields like healthcare and autonomous driving, where decisions must be clear.

The method uses concept bottleneck models, which enable AI to base its predictions on human-understandable concepts. Traditional approaches rely on expert-defined concepts that can be incomplete or ill-suited, sometimes lowering model performance.

Researchers instead created a system that extracts concepts the AI learned during training. A sparse autoencoder selects key features, and a multimodal language model turns them into plain-language descriptions and labels.

The resulting module forces the AI to make predictions using only those extracted concepts.

Tests on bird classification and medical image datasets showed that the new method improved accuracy and provided clearer explanations. Findings suggest that using a model’s internal concepts can boost transparency and accountability in AI systems.

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

Lenovo introduces rollable laptop and AI agent

Redefining how people interact with technology, Lenovo is advancing through rollable laptops, foldable devices and adaptive AI systems that anticipate user needs.

The company is shifting from manufacturing hardware to creating multi-platform systems that adapt seamlessly to workflows instead of relying solely on traditional devices.

Qira, Lenovo’s personal AI super-agent, transfers tasks across devices while maintaining context and history with user permission. It can suggest actions and predict needs, aiming to improve productivity and employee satisfaction, although security and privacy concerns remain significant.

The rollable laptop features a 14-inch screen that expands vertically to 16.7 inches, providing immersive experiences for gaming and content consumption while remaining portable.

Lenovo is also exploring voice-driven tools, including AI Workmate prototypes, allowing users to create presentations and digital content simply through speech.

By combining innovative screen designs with intelligent AI agents, Lenovo aims to create unified ecosystems that prioritise user experience and adaptability instead of focusing solely on device specifications.

The company believes these technologies will gradually become culturally accepted, similar to self-driving cars.

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

EU considers stronger child protection in Digital Fairness Act

Capitals across the EU are being asked to discuss how stronger child protection measures should be incorporated into the upcoming Digital Fairness Act (DFA).

The initiative comes as policymakers attempt to address growing concerns about how online platforms expose minors to harmful content, manipulative design practices, and unsafe digital environments.

According to a document circulated during Cyprus’s Council presidency of the European Union, member states are expected to debate which concrete safeguards should be introduced as part of the broader consumer protection framework.

Officials are exploring whether new rules should require platforms to adopt stricter safeguards when designing digital services used by children.

The discussions are part of the European Union’s broader effort to strengthen digital governance and consumer protection across online platforms. Policymakers are increasingly focusing on how platform design, recommendation algorithms, and monetisation models may affect younger users.

The proposals could complement existing EU regulations targeting large digital platforms, while expanding protections specifically focused on minors.

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

Australia introduces strict online child safety rules covering AI chatbots

New Age-Restricted Material Codes have begun to be enforced in Australia, requiring online platforms to introduce stronger protections to prevent children from accessing harmful digital content.

The rules apply across a wide range of services, including social media, app stores, gaming platforms, search engines, pornography websites, and AI chatbots.

Under the framework, companies must implement age-assurance systems before allowing access to content involving pornography, high-impact violence, self-harm material, or other age-restricted topics.

These measures also extend to AI companions and chatbots, which must prevent sexually explicit or self-harm-related conversations with minors.

The rules form part of Australia’s broader online safety framework overseen by the eSafety Commissioner, which will monitor compliance and enforce the codes.

Companies that fail to comply may face penalties of up to $49.5 million per breach.

The policy aims to shift responsibility toward technology companies by requiring them to build protections directly into their platforms.

Officials in Australia argue the measures mirror long-standing offline safeguards designed to prevent children from accessing adult environments or harmful material.

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

AI legal advice case asks whether ChatGPT crosses legal boundaries

A newly filed lawsuit against OpenAI raises a key issue: Does allowing generative AI systems like ChatGPT to provide legal advice violate laws that bar the unauthorised practice of law (UPL)? UPL means providing legal services, such as drafting filings or giving advice, without the required legal qualifications or a state licence.

The case claims an individual used ChatGPT to prepare legal filings in a dispute with Nippon Life Insurance, prompting the company to argue OpenAI should be held responsible for the outcome.

The lawsuit claims ChatGPT helped the user challenge a settled legal dispute. As a result, the company had to spend additional time and resources responding to filings produced with ChatGPT. The claim alleges tortious interference with a contract, which is the unlawful disruption of an existing agreement between two parties by causing one of the parties to breach or alter it.

Ultimately, this disrupted another party’s contractual relationship. The suit also claims unauthorised practice of law and abuse of the judicial process, which means using the legal system improperly to gain an advantage. It argues OpenAI should be liable because ChatGPT operates under its control. The dispute centres on whether AI systems should analyse disputes and offer legal advice like a lawyer.

Advocates argue the tools could widen access to legal advice. They could make legal support more accessible and affordable for those who cannot easily hire a lawyer. However, US legal frameworks restrict the provision of legal advice to licensed lawyers. The rules are designed to protect consumers and ensure professional accountability.

Critics argue that limiting legal advice to licensed lawyers preserves an expensive monopoly and hinders access to justice. AI-driven legal tools highlight this tension over the future of legal services.

The outcome of this lawsuit will likely hinge on whether AI-generated responses constitute intentional legal advice and if OpenAI can be held liable for such outputs. Even if it fails, the case foregrounds the broader debate about granting generative AI a legitimate role in legal guidance.

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

ChatGPT ‘adult mode’ launch delayed as OpenAI focuses on core improvements

OpenAI has postponed the launch of ChatGPT’s ‘adult mode’, a feature designed to let verified adult users access erotica and other mature content.

Teams are focusing on improving intelligence, personality and proactive behaviour instead of releasing the feature immediately.

A feature that was first announced by Sam Altman in October, with an initial December rollout, aiming to allow adults more freedom while maintaining safety for younger users.

The project faced an earlier delay as internal teams prioritised the core ChatGPT experience.

OpenAI stated it still supports the principle of treating adults like adults but warned that achieving the right experience will require more time. No new release date has been provided.

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