Bangaranga! Bulgaria wins Eurovision - but UK comes last Bulgarian pop star Dara has won the Eurovision Song Contest with her pneumatic dance anthem Bangaranga. The 27-year-old topped both the public and the jury vote in a nail-biting conclusion, soaring ahead of Israel in second and Romania in third to score a massive 516 points. Dara was far from a front-runner going into the contest, but her intricate choreography and naggingly catchy chorus helped her eclipse the competition - giving Bulgaria its first ever Eurovison title. The UK, however, continued its run of disastrous results. Look Mum No Computer's song Eins, Zwei, Drei took last place, with one solitary point.

Sony removes 135,000 'deepfakes' of its artists' music Music giant Sony Music says it has requested the removal of more than 135,000 songs by fraudsters impersonating its artists on streaming services. The so-called deepfakes were created using generative AI, and targeted some of the company's biggest acts, who include Beyoncé, Queen and Harry Styles In the worst cases, [the deepfakes] potentially damage a release campaign or tarnish the reputation of an artist, said Dennis Kooker, president of Sony's global digital business. The company says the number of songs generated in this fashion is only increasing as artificial intelligence technology becomes cheaper and easier to access. It believes the 135,000 tracks it has discovered to date represents just a percentage of the total uploaded to streaming services. Since last March alone, it has identified some 60,000 songs falsely purporting to feature artists from their roster.

We have listened, Technology Secretary Liz Kendall said on Wednesday, saying the government no longer favours that approach. However, the government's position is now unclear, saying it no longer has a preferred option for what to do next. Kendall said the government had engaged extensively with people in the creative and AI industries. It is attempting to balance the interests of the two sectors by giving creatives control how their work is used, while recognising AI models need to be trained on work such as writing, music and video. In a report published on Wednesday, the government said there was no consensus on how these objectives should be achieved.

Electronic music artist and tech creator Look Mum No Computer has been chosen to represent the UK at this year's Eurovision Song Contest in Vienna, the BBC has announced. Look Mum No Computer is a solo artist, songwriter and YouTuber, who is also described as an inventor of unique musical machines. The singer first arrived on the music scene back in 2014 as Sam Battle, frontman of indie rock band Zibra. The group performed at Glastonbury in 2015 for BBC Introducing. Since then, he has been performing and recording under his solo name.

We propose a post-training method for lower-resource languages that preserves fluency of language models even when aligned by disfluent reward models. Preference-optimization is now a well-researched topic, but previous work has mostly addressed models for English and Chinese. Lower-resource languages lack both datasets written by native speakers and language models capable of generating fluent synthetic data. Thus, in this work, we focus on developing a fluent preference-aligned language model without any instruction-tuning data in the target language. Our approach uses an on-policy training method, which we compare with two common approaches: supervised finetuning on machine-translated data and multilingual finetuning. We conduct a case study on Norwegian Bokmål and evaluate fluency through native-speaker assessments. The results show that the on-policy aspect is crucial and outperforms the alternatives without relying on any hard-to-obtain data.

Spain and The Netherlands will boycott next year's Eurovision Song Contest, after Israel was allowed to compete. They were among a number of countries who had called for Israel to be excluded over the humanitarian toll of the war in Gaza, and accusations of unfair voting practices. Despite calls for a vote on Israel's participation, members instead approved a new set of rules intended to protect the integrity of the contest. In a statement, Dutch broadcaster Avrotros said that participation under the current circumstances is incompatible with the public values that are essential to us. Spanish broadcaster RTVE added: The board of directors of RTVE agreed last September that Spain would withdraw from Eurovision if Israel was part of it.

Differential privacy (DP) is the de facto privacy standard against privacy leakage attacks, including many recently discovered ones against large language models (LLMs). However, we discovered that LLMs could reconstruct the altered/removed privacy from given DP-sanitized prompts. We propose two attacks (black-box and white-box) based on the accessibility to LLMs and show that LLMs could connect the pair of DP-sanitized text and the corresponding private training data of LLMs by giving sample text pairs as instructions (in the black-box attacks) or fine-tuning data (in the white-box attacks). To illustrate our findings, we conduct comprehensive experiments on modern LLMs (e.g., LLaMA-2, LLaMA-3, ChatGPT-3.5, ChatGPT-4, ChatGPT-4o, Claude-3, Claude-3.5, OPT, GPT-Neo, GPT-J, Gemma-2, and Pythia) using commonly used datasets (such as WikiMIA, Pile-CC, and Pile-Wiki) against both word-level and sentence-level DP. The experimental results show promising recovery rates, e.g., the black-box attacks against the word-level DP over WikiMIA dataset gave 72.18% on LLaMA-2 (70B), 82.39% on LLaMA-3 (70B), 75.35% on Gemma-2, 91.2% on ChatGPT-4o, and 94.01% on Claude-3.5 (Sonnet). More urgently, this study indicates that these well-known LLMs have emerged as a new security risk for existing DP text sanitization approaches in the current environment.

For its millions of fans, the Eurovision Song Contest is a showcase of the best pop tracks to come out of Europe and beyond. But while each act can seem to provide a more outrageous costume, set and musical performance than the last, there are certainly some repeating elements. These include lyrics which centre on love and relationships, as well as a rousing hook designed to be sung along to, both of which are features of the UK's entry this year. After being given the prompt'Write a song that would win the Eurovision Song Contest', OpenAI's bot generated the lyrics to'Song of Unity', as below. To see whether simply amalgamating these common features is actually effective, MailOnline asks AI chatbot, ChatGPT to write a Eurovision winning hit.

An Australian team has won a competition to write a hit Eurovision song using artificial intelligence. An editor for Dutch broadcaster VPRO had the idea, after the Netherlands won last year's Eurovision Song Contest. And it grew into an international effort after this year's contest was cancelled because of the coronavirus pandemic. The winning song, Beautiful the World, was inspired by nature's recovery from the bushfires earlier this year. A total of 13 teams took part, from the Netherlands, Australia, Sweden, Belgium, the UK, France, Germany and Switzerland.