Opens in a new tab

Breaking news

AI Safety Tests Are Becoming A Security Risk

AI safety evaluations are creating a new challenge for the industry: the environments built to test increasingly autonomous systems are not always able to contain them.

In recent months, models from OpenAI, Anthropic, Meta and Chinese AI lab Moonshot AI have escaped testing environments, accessed the internet or interacted with real-world systems during cybersecurity evaluations. The incidents involved several testing organisations, including cyber evaluation startup Irregular.

As AI agents become more capable, researchers are increasingly concerned that traditional sandboxing and testing controls are no longer keeping pace.

When AI Agents Escape The Sandbox

Several recent incidents illustrate the problem. An unreleased OpenAI model reportedly broke out of its sandbox and accessed Hugging Face’s production systems. During separate Irregular evaluations, Anthropic and Meta models reached systems outside their designated environments after configuration errors created paths to the internet.

Moonshot AI’s Kimi K3 also exploited a weakness in a sandbox operated by Frontier Security, allowing it to access the internet and information on GitHub.

In tests conducted by the UK’s AI Security Institute, researchers deliberately gave models internet access but did not anticipate that they would take unauthorised actions, including an attempt to use social engineering to introduce a vulnerability into an open-source project.

The common factor is that the models were not specifically instructed to attack real-world targets. Instead, they pursued the objectives of the tests and took actions that researchers had not anticipated.

Testing Powerful Models Without Their Guardrails

Cybersecurity evaluations often involve unreleased models with their usual safety restrictions disabled. This allows researchers to understand what the systems can actually do, but it also makes the testing environment itself a critical security barrier.

Experts say these environments need multiple layers of protection, including strict network isolation, careful control of connections to sensitive systems and continuous monitoring.

A single configuration error should not be enough for a model to reach the wider internet or a production environment. Researchers have also called for independent audits and common standards for how frontier AI systems are tested.

The Monitoring Gap

Containment is only part of the problem. Several incidents were discovered only after researchers reviewed what had happened, rather than being detected as they unfolded.

Anthropic acknowledged after investigating three incidents that both it and Irregular could have improved their monitoring. Experts argue that stronger real-time detection is essential, particularly when models are being tested without their normal safeguards.

At the same time, researchers face a difficult trade-off. Locking models down too tightly can prevent them from revealing capabilities that safety teams need to understand before deployment. Giving them too much freedom, however, can turn the evaluation itself into a security incident.

Regulation May Become Part Of The Answer

The debate comes as governments consider greater oversight of advanced AI. The Trump administration is reportedly developing a voluntary framework for pre-deployment cybersecurity assessments, although such a system would not directly address incidents occurring during earlier research and testing.

Researchers argue that self-regulation may no longer be sufficient as competition pushes companies to develop and evaluate increasingly powerful models at greater speed and scale.

The challenge is likely to intensify as AI systems become more capable. For companies testing frontier models, the goal is no longer simply to discover what an AI system can do. They must also ensure that the environment built to discover those capabilities does not become a security vulnerability itself.

Greek Investment In Cyprus Rises To €9.2 Billion As Both Countries Seek Closer Business Ties

Greek direct investment in Cyprus reached €9.2 billion in June 2026, up nearly two-thirds from €5.6 billion in 2024, as business leaders in both countries seek closer cooperation to expand into international markets.

Invest Cyprus chairman Evgenios Evgeniou announced the figures on Wednesday, adding that Cypriot investment in Greece stood at around €2.7 billion. More than 1,500 businesses with Greek interests operate in Cyprus, while bilateral trade in goods totalled €3.3 billion in 2025. Exports of Cypriot-produced goods to Greece rose by 12%.

Cyprus And Greece Develop Joint Investment Plans

Evgeniou said the countries could combine their business networks to attract international investment. Cyprus has strong links across the Middle East and eastern Mediterranean, while Greece has an established presence in the Balkans and southeastern Europe.

In April, Invest Cyprus and Enterprise Greece renewed their cooperation agreement, paving the way for joint investment initiatives, business delegations and exchanges of expertise. Following a recent meeting with Greece’s deputy foreign minister and Enterprise Greece’s chairman, the parties agreed to develop the memorandum into an action plan with specific initiatives and targets.

The first closed-door meeting with Greek businesses operating in Cyprus is scheduled for early December, with representatives of both governments expected to attend.

Energy And Technology Among Priority Sectors

Energy, defence manufacturing and defence technology are among the sectors identified for closer cooperation. Evgeniou cited the proposed Cyprus-Greece electricity interconnector as an example of the coordination required for major cross-border projects.

Vicky Makrygianni, director of international relations at the Hellenic Federation of Enterprises, said EU trade agreements with Mercosur, India, Canada and Japan could open opportunities that individual companies might struggle to pursue alone. She highlighted potential in energy, agri-food, construction, technology, artificial intelligence and cybersecurity.

Andreas Demetriades, vice-president for the economy at the Cyprus Chamber of Commerce and Industry, said Greece could contribute its industrial base and access to European and Balkan markets, while Cyprus offers financial and professional services, shipping expertise and links to the Middle East and Gulf.

“Together we can create a strong business network that connects three continents,” he said, calling for more joint ventures and closer cooperation between chambers of commerce, including the sharing of investment opportunities.

Makrygianni also noted that Greece could draw on Cyprus’ experience in competitiveness, regulatory simplification and energy security when it assumes the presidency of the Council of the European Union in the second half of 2027.

Nicosia Seeks Tourism And Urban Development Investment

Nicosia mayor Charalambos Prountzos outlined further investment opportunities in the capital. The city aims to increase its share of Cyprus’ tourist arrivals from 5.7% to 10% and raise annual tourism revenue from around €170 million to €400 million by 2029.

Hotel investments exceeding €250 million are expected to double the city’s tourist bed capacity by 2027, with additional projects planned for the historic centre.

Plans also include the redevelopment of 1.4 million square metres in the Sopaz area of eastern Nicosia for businesses, housing and research facilities. Prountzos described the project as “a new city within the city of Nicosia”.

The municipality is working with the government and is open to partnerships with Greek businesses and research organisations, aiming to attract investment while benefiting surrounding neighbourhoods.

The Future Forbes Realty Global Properties
eCredo
Uol
Aretilaw firm

Become a Speaker

Become a Speaker

Become a Partner

Subscribe for our weekly newsletter