Yazılar

MI5 chief warns AI could pose future security risks, but dismisses “Hollywood doom”

/in /tarafından

The head of Britain’s domestic intelligence agency, MI5, has warned that artificial intelligence systems acting independently of human oversight could one day pose serious national security challenges — though he dismissed notions of a “Terminator”-style apocalypse.

In his annual speech on national threats, MI5 Director General Ken McCallum said that while AI is already being used to strengthen British security operations, it is also being exploited by terrorists, hostile states, and cybercriminals. He said AI tools are helping adversaries spread propaganda, conduct reconnaissance, and manipulate elections.

“But in 2025, while contending with today’s threats, we also need to scope out the next frontier: potential future risks from non-human, autonomous AI systems which may evade human oversight and control,” McCallum said.

He emphasized that his warning was not a prediction of science-fiction-style catastrophe, but a call for preparedness as AI technology rapidly evolves. “Given the risk of hype and scare-mongering, I will choose my words carefully. I am not forecasting Hollywood movie scenarios,” he noted.

McCallum added that while AI may never intentionally mean harm, ignoring its potential dangers would be “reckless.” MI5 and other intelligence agencies are studying the long-term implications of increasingly autonomous systems.

The remarks reflect a broader debate within global intelligence and tech circles about balancing the benefits of AI innovation with the risks of automation and loss of control over powerful systems.

Nabla Bio and Takeda deepen AI-powered drug discovery partnership

/in /tarafından

U.S. biotech startup Nabla Bio has expanded its collaboration with Japanese pharmaceutical giant Takeda, signing a new multi-year agreement to further integrate artificial intelligence (AI) into the drug discovery process.

The partnership, building on an initial 2022 deal, includes upfront and research payments in the double-digit millions, with Nabla eligible for success-based milestones exceeding $1 billion. The companies will use Nabla’s proprietary Joint Atomic Model (JAM) platform to design next-generation protein-based therapeutics, including multi-specific drugs targeting hard-to-treat diseases.

Nabla CEO Surge Biswas described JAM as a system that “responds to molecular queries the way ChatGPT answers text questions,” generating custom antibody designs that meet specific biological targets. The company says its technology offers one of the fastest feedback loops in the biotech sector — only three to four weeks from computational design to laboratory testing.

Takeda’s renewed focus on scalable, AI-driven drug types follows its decision to exit cell therapy research earlier this year. The Japanese firm also recently joined a consortium with Bristol Myers Squibb to develop AI models using shared pharmaceutical data.

Nabla expects to deliver first-in-human data from its AI-designed therapeutics within one to two years, marking another milestone in the industry’s growing reliance on AI to cut development timelines and boost innovation.

Taiwan warns of surge in Chinese cyberattacks and “online troll army”

/in /tarafından

Taiwan’s National Security Bureau (NSB) has reported a 17% rise in Chinese cyberattacks targeting its government systems so far in 2025, amounting to an average of 2.8 million attacks per day. The agency warned that Beijing is deploying an “online troll army” to amplify disinformation and undermine public trust in the island’s institutions.

The NSB report, presented to parliament, described these as systemic cyberattacks focusing on key sectors such as defence, telecommunications, energy, and healthcare. Beyond espionage, the operations reportedly use the dark web, internet forums, and social media to spread fabricated content and erode public confidence in Taiwan’s cybersecurity capabilities.

Taiwan’s authorities accuse China of using “grey-zone” tactics — hybrid measures that combine military drills, cyber intrusions, and propaganda — to pressure the island into accepting Beijing’s sovereignty claims. China’s Taiwan Affairs Office declined to comment, though Beijing has repeatedly denied involvement in hacking and instead claims it is a victim of Taiwanese cyber operations.

The report also flagged more than 10,000 suspicious social media accounts, mostly on Facebook, that collectively spread over 1.5 million pieces of disinformation. These campaigns reportedly promote pro-China narratives, attack Taiwan’s leadership, and attempt to sow distrust toward the United States, Taipei’s key ally and arms supplier.

According to the NSB, AI-generated memes and targeted digital propaganda have become central tools in China’s information warfare strategy ahead of Taiwan’s elections and trade talks with Washington.