Anthropic's Settlement Rejected, Google Becoming the Instagram of AI, Mistral's Series C, Sam Altman's "Dead Internet" Concerns
Federal judge blast's Anthropic's copyright settlement, Google is beating OpenAI in AI video, Mistral's Series C shows Europe is behind in AI, and Sam Altman worries about the "dead internet theory"
⚖️ Anthropic's $1.5B Settlement Rejected, Judge Cites Incomplete Details
A federal judge has rejected a record-breaking $1.5 billion settlement agreement in the copyright lawsuit filed by authors against AI company Anthropic. Judge William Alsup's decision stems from concerns that the deal was rushed and incomplete, potentially short-changing the very authors it aims to compensate.
Judge Alsup expressed an "uneasy feeling" about the settlement, particularly the significant amount of money at stake and the potential for class members to get "the shaft." He was reportedly misled by the initial deal and found it "nowhere close to complete," lamenting that "important questions" were left unanswered, including:
A definitive list of the 500,000 authors involved.
A precise list of the pirated works used to train Anthropic's large language models.
A clear process for notifying class members about the settlement.
A user-friendly claim form that allows authors to easily opt-in or opt-out.
While the lawyers representing the authors called the deal a "landmark settlement" that "far surpasses any other known copyright recovery," the judge was not convinced. Alsup's primary worry is that once the lawyers secure the settlement, they may lose interest, leaving authors with a raw deal.
Judge Alsup has given the lawyers a tight deadline to iron out the details. They have until September 15 to submit a final list of the works involved.
Furthermore, the lists of works and class members, along with the claim form, must be reviewed and approved by the court by October 10 before any preliminary approval can be granted. The judge also mandated that the lawyers must create a clear and effective notice for all class members, ensuring they fully understand their options to participate in or withdraw from the settlement.
The settlement, which aims to pay ~$3,000 per pirated work, now hangs in the balance, creating a critical hurdle for what was seen as a landmark resolution.
🤖 Google's Veo 3 Gets a Pro-Grade Update, the Instagram of AI?
Google has enhanced its Veo 3 AI video generator with key upgrades, making it more appealing to professional creators. The update adds 1080p resolution and 9:16 vertical formats, along with halving the price to just $0.15 per second.
With these latest updates, Google is positioning itself to become the "Instagram of AI." By focusing on fast, affordable, and high-quality video generation tailored for social media, Google is directly targeting the viral, short-form content market.
The new 1080p resolution and vertical formats, combined with a low price of $0.15 per second, make it the ideal tool for creators aiming for high-volume iteration.
If Google succeeds in winning video, their strategy could make OpenAI's ChatGPT, the "Twitter of AI." While ChatGPT is taking the market by storm, the interface is still primarily text-based, as Twitter was back went it went viral in 2007.
We wouldn’t know then, but Instagram and Facebook would end up dominating the social media market with their image and video offerings. Google is putting itself in the pole position if a similar dynamic plays out with consumer AI.
💰 Mistral AI Closes a $1.9B Round at an $12.8B Valuation
Paris-based Mistral AI raised a $1.9 billion Series C round, more than doubling their valuation to $12.8 billion. The round was led by Extreme Ultraviolet (EUV) lithography leader ASML, with prior investors General Catalyst and Andreessen Horowitz participating.
This capital influx cements Mistral as Europe's leading AI company and will be used to fuel its open-source models and a new AI data center project. This also highlights how far behind Europe is, given Mistral’s US-based peers have 10x their valuation (OpenAI $500 billion, Anthropic $183 billion).
The investment validates Mistral’s strategic focus on developing efficient, open-source models as a counterpoint to the closed systems of American giants like OpenAI. With a valuation that outpaces many rivals, Mistral's success underscores the global compute race and Europe's ambition to become a major player in AI.
The partnership with ASML is particularly notable, with a goal to integrate AI models into ASML's semiconductor equipment, research, development, and operations to accelerate innovation and deliver higher-performance lithography systems with faster time-to-market for customers.
This collaboration also strengthens European technological sovereignty by combining ASML's manufacturing expertise with Mistral's AI capabilities.
💀 Sam Altman's "Dead Internet" Warning
OpenAI CEO Sam Altman recently reignited the "Dead Internet Theory," warning on X that the rise of AI-generated "slop" is making online content feel less human and authentic. He noted that even real human posts are beginning to mimic the style of large language models.
Altman's warning, coming from the epicenter of the generative AI boom, underscores a growing fear that AI is polluting the digital commons. Some estimates suggest that as much as 90% of web text could be synthetic by 2026. This poses significant risks to trust in online information, from social media to search results.
This observation highlights the double-edged sword of AI: it empowers creation but can also degrade the very spaces where knowledge is shared. While solutions like watermarking and detection technology are being explored, the question remains whether the degradation of online content is an inevitable consequence of AI's rapid growth.
And if it is, where will AI companies get their information in the future?