Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
€19.00Add to cart
€19.00Add to cart
Since 2 August, new transparency provisions under the European Union’s AI Act have begun applying, introducing rules intended to make it easier to recognise when AI is talking to us and when what we are seeing, hearing or reading has been artificially created or manipulated.
For years, the internet has operated on a fairly simple assumption: unless there was reason to suspect otherwise, the photograph, voice or article in front of you was probably created by a human being.
Artificial intelligence has made that assumption increasingly unreliable.
Since 2 August, new transparency provisions under the European Union’s AI Act have begun applying, introducing rules intended to make it easier to recognise when AI is talking to us and when what we are seeing, hearing or reading has been artificially created or manipulated.
The principle is simple. If technology is becoming extraordinarily good at imitating reality, the technology should also help us identify the imitation.
The rules do not mean that every business using artificial intelligence must attach an enormous AI warning to everything it produces.
Instead, Article 50 creates different obligations for different uses.
Providers of generative AI systems must ensure that AI-generated or manipulated audio, images, video and text can be identified in a machine-readable form. That could involve a “label” created by metadata, digital watermarks or other technical marking systems designed to survive as content travels across the internet.
There are additional obligations for those deploying AI.
People interacting directly with certain AI systems, such as chatbots, generally need to be told that they are dealing with a machine rather than a human.
Deepfakes must also be disclosed as artificially generated or manipulated.
And there is a particularly interesting rule for publishers: AI-generated or manipulated text intended to inform the public about matters of public interest must be identified as such, unless it has undergone human review or editorial control and a person or organisation assumes editorial responsibility for its publication.
That distinction matters enormously for newspapers, magazines and other professional publishers. Using AI somewhere in the production process does not automatically turn an article into something requiring an “AI-generated” label.
Brussels has also recognised that context matters.
Artistic, creative, satirical and fictional works receive different treatment. Where a deepfake forms part of an obviously creative work, for example, disclosure requirements are designed not to interfere unnecessarily with the enjoyment of the work.
The objective is therefore less about putting warning signs on artificial intelligence than preventing people from being deceived by it.
That becomes increasingly important as synthetic voices become indistinguishable from real ones and fabricated photographs and videos can depict events that never occurred.
The European Commission says the rules are intended to reduce precisely these risks of deception and manipulation.
The EU has also published a voluntary Code of Practice showing AI providers and users how they can comply, alongside guidelines explaining the new obligations.
And there are teeth behind the legislation. Breaches of the AI Act can, depending on the provision and circumstances, expose companies to substantial penalties. For infringements falling within the relevant penalty tier, fines can reach €15 million or 3% of worldwide annual turnover.
Perhaps the most significant part of the new regime, however, is not the fine.
It is the change in expectations.
Photography once provided evidence that something had happened. A recording suggested somebody had actually said something. A written article at least implied that somebody had written it.
AI has weakened all three assumptions.
Europe’s answer is not to ban synthetic content. It is to insist that, when artificial intelligence becomes good enough to imitate reality, the audience should still have a reasonable chance of knowing the difference.
