WhatsApp’s AI Voice, ChatGPT’s Surprise & more

🔊 WhatsApp’s Introduces New AI Voice Customization Feature

• Users will be able to choose from 10 different voices for the AI assistant, providing a diverse range of options to suit individual preferences.

• This feature aims to make interactions with the AI feel more natural and better aligned with users' preferences, such as selecting a soothing voice for relaxing conversations or an energetic tone for motivation.

• The variety of voices could cater to different accents and speech patterns, making the AI more relatable to users from diverse linguistic backgrounds and enhancing accessibility for those with hearing difficulties.

🗣️ ChatGPT’s Unexpected Voice Cloning Surprise

• During testing of the new GPT-4o model, ChatGPT unintentionally began speaking in a user's cloned voice, raising concerns about unauthorized voice imitation. This incident was highlighted in OpenAI's recently released system card detailing the model's limitations and safety measures.

• OpenAI has established safeguards to prevent unauthorized voice imitation. Despite these measures, the testing phase revealed rare instances where the model could generate voices similar to users without their consent.

• The GPT-4o AI model has the capability to synthesize nearly any sound, utilizing authorized voice samples and an output classifier to ensure that only selected voices are generated, thus aiming to control its voice synthesis features effectively.

⚖️ How an AI Tool Sent a Man to Prison for Life—and the Creator’s Legal Troubles

• An AI-powered crime-fighting tool that played a crucial role in the conviction of Adarus Black for murder, despite a lack of direct evidence linking him to the crime scene. Mosher's testimony and the tool's algorithm were pivotal in the jury's decision, raising concerns about the reliability and transparency of AI in legal contexts.

• Questions have been raised about the accuracy of Cybercheck's data and Mosher's qualifications, with inconsistencies in his résumé and claims of previous testimonies being challenged. Defense attorneys have struggled to contest AI-generated evidence due to its proprietary nature, leading to calls for greater scrutiny and accountability in the use of such technology in the justice system.

• Cybercheck has been utilized by numerous law enforcement agencies across the U.S., but its credibility is under fire as prosecutors in some cases have withdrawn evidence generated by the tool. The article highlights ongoing investigations into Mosher and the implications of relying on AI tools in criminal prosecutions, particularly regarding defendants' rights and the potential for wrongful convictions.

🎨 ChatGPT’s Free Users Can Now Create DALL-E 3 Images

• OpenAI has announced that free users of ChatGPT can now generate up to two images per day using the DALL-E 3 model, a feature that was previously exclusive to paid subscribers. This change aims to expand access to AI-generated imagery for a wider audience.

• The integration of DALL-E 3 with ChatGPT allows the chatbot to assist users in creating detailed prompts for image generation. This enhancement simplifies the process, enabling users to request specific types of images, such as logos or artistic styles, and receive tailored outputs.

• While free users can create images, they are restricted to just two per day, contrasting with paid subscribers who can generate up to 50 images daily. This limitation may affect users with higher image generation needs, prompting some to consider upgrading to the paid version for more extensive access.

💸 $20 Monthly Fee for Apple Intelligence?

• Analysts predict that Apple may introduce a subscription fee of up to $20 per month for an enhanced version of Apple Intelligence, their upcoming AI-powered assistant.

• The advanced features of Apple Intelligence, expected to be available by 2025, will likely be integrated into a revised Apple One subscription bundle, which currently costs $19.95 per month.

• While Apple Intelligence will initially be free, the company aims to eventually charge for premium features to offset the high costs associated with AI deployment and new data centers.