MPs have cautioned that the UK's ability to be at the forefront of AI regulation will be jeopardised unless new legislation is enacted in November.
Ministers on the Commons Technology Committee cautioned that the EU could surpass the UK in attempts to ensure AI security, unless steps are taken.
At the beginning of November, the UK will be hosting an international summit regarding Artificial Intelligence.
The government informed the BBC they are open to taking additional action if necessary.
However, the company did not articulate whether they endorsed the idea of formulating a law urgently. Instead, a spokesperson spotlighted the summit and the initial £100 million investment to form a task-force devoted to promoting the secure utilisation of AI frameworks.
The UK government declared that it has allocated more funding towards AI safety than any other government globally.
It is stated in a report which was released on Thursday that if no legislation is brought forth in the King's Speech on 7 November, the soonest it could potentially be enacted would be by 2025.
The report cautions that delaying legislation for a period of two years might mean the UK fails to keep up with other countries, such as the EU, that have already implemented AI regulations, and that these standards might be difficult to replace.
The report proposes that the circumstance could be analogous to regulations regarding data protection, where British laws have followed the regulation set by the European Union.
Despite the acknowledgement in the government's white paper on AI regulation that legislation may be necessary in the future, Rishi Sunak has suggested that to begin with, "we can probably do lots of this without legislation".
A major element of his proposal is the November summit that the state states will be the inaugural global conference about AI safety.
The committee maintained that as many nations as feasible should be asked to attend, with China among them.
The report puts forward twelve "challenges" that the UK government must tackle, including:
Employing copyrighted works to coach Artificial Intelligence (AI) systems poses a challenge.
Generative AI systems have been developed to the point where they can now manufacture artistic, acting and musical works in the manner of renowned creators.
In order to achieve this accomplishment, AI is trained on sizable amounts of copyrighted material. Numerous authors, performers, artists, and musicians contend that their works should not be trained on AI without obtaining permission and paying them accordingly.
A voluntary arrangement is being formulated that will enable AI firms to access copyrighted works and still provide benefits to the artists, as the report observes.
In February, the government decided against providing a copyright exception for AI companies.
MPs have warned that the capability of AI to mimic humans could lead to fraudulent activities such as the diffusion of false information, financial scams or the deception of bank security systems which rely on voice recognition.
Yesterday, the National Cyber Security Centre warned that currently, there are no failsafe measures that prevent certain attacks from occurring on large language models, which are utilized in many chatbots. They added that the models are not adequately protected against these malicious actions.
MPs generally agreed with the government's stance on keeping AI secure, which avoids creating a new AI governing body, but instead entrusts control to the existing regulators based on the tasks that the AI carries out.
Hugh Milward of Microsoft UK told the committee that, when compared to the EU's approach, he preferred this approach; he described the EU's strategy as "a model of how not to do it".
However, he has previously warned the BBC to be cautious when introducing UK laws as well. He has cautioned that attempting to accomplish too much in a single legislation could lead to unwanted additions, like a Christmas tree that everyone decorates with their own ornaments.
Kommentare