Newsroom
A new form of deepfake technology, known as voice cloning, has recently surfaced, raising concerns about its potential misuse. OpenAI, a prominent tech company, has developed voice cloning software capable of replicating anyone's voice within just 15 seconds of recorded speech. However, the company has opted not to release the technology to the public due to its perceived risks, especially in the context of an election year. Instances of AI-generated voices being used in robocalls have prompted investigations by authorities, highlighting the potential for malicious exploitation.
Meanwhile, in a landmark development, the US and UK have signed an unprecedented agreement focusing on artificial intelligence (AI) safety. This historic pact marks the first formal collaboration between nations to jointly test and assess risks associated with emerging AI models. The agreement, signed by UK Science Minister Michelle Donelan and US Commerce Secretary Gina Raimondo, outlines plans for the two governments to share technical knowledge, information, and talent in the realm of AI safety.
Under the agreement, the UK's AI Safety Institute (AISI) and its US counterpart will exchange expertise through researcher secondments and collaborate on the evaluation of private AI models. The partnership aims to address concerns surrounding the potential misuse of AI, including national security threats and broader societal impacts.
The UK's AISI, backed by government support and chaired by tech investor Ian Hogarth, has already begun testing existing and unreleased AI models, with a focus on cybersecurity risks. Tech giants like OpenAI, Google DeepMind, Microsoft, and Meta have voluntarily committed to subjecting their latest AI models to scrutiny by the AISI.
Source: Financial Times and Daily Mail UK
