A German court's decision regarding Alphabet (GOOGL) over the past two days has captured significant attention. This is not the typical, reflexive ruling against a U.S. tech giant often seen in Europe. The judgment is grounded in logical reasoning and is poised to have a profound impact on the entire artificial intelligence sector. A German court has ruled that Google must bear legal responsibility for statements made in AI-generated answers within its search results. Google stated the ruling is currently preliminary, but it directly addresses a critical unresolved question: who is liable when an AI model outputs costly misinformation?
Key Details of the Ruling
Two companies sued Google after its AI Overview feature incorrectly described them as fraudulent entities. The court ultimately sided with the plaintiffs (the German publication *Decoder* first reported the verdict). The judgment notes that these inaccuracies originated entirely from Google's AI model: the content generated by AI Overview included assertions not present in the source webpages it cited. The ruling states: "These statements are unique assertions generated by Google's AI tool, for which Google must assume responsibility."
This verdict also highlights a crucial legal conflict: the protections afforded to U.S. tech companies like Google under Section 230 of the U.S. Communications Decency Act may not apply to AI-generated content. This law stipulates that platforms are not liable for third-party content on their services. The ruling also mentions that similar protective rules exist locally in Germany: "For traditional search engines, courts typically rule that operators are not required to verify every linked piece of content in advance."
However, the judgment clearly distinguishes the fundamental difference: traditional search results merely list external links, while AI Overview synthesizes fragmented information to generate entirely new, original text. Google's official response stated that the vast majority of AI Overview answers are accurate, but it acknowledged the feature occasionally produces errors.
Google's defense in court argued that users could verify information by clicking the source links attached to the AI answers. The court countered this: "If the law were to consider AI Overview text completely untrustworthy, requiring users to verify every link, the very purpose of this feature would be entirely negated." This perspective is highly persuasive.
Oracle's Mixed Financial Results
Oracle (ORCL) reported that the expansion of its AI cloud business met expectations, but persistent weakness in its traditional software segment is dragging down the company's overall growth rate. Software was long Oracle's core pillar, but growth in its AI data center business has now overtaken it.
Oracle released its fiscal fourth-quarter (ending May) results on Wednesday: total revenue increased 21% year-over-year, a deceleration of 1 percentage point from the previous quarter. The company attributed the slowdown primarily to foreign exchange fluctuations, stating that growth actually accelerated on a constant-currency basis.
Even though the overall 21% revenue growth appears solid, it fails to capture the explosive growth in the cloud segment: cloud revenue surged 93% year-over-year this quarter to $5.8 billion.
The software business presents a stark contrast. As enterprise customers shift from perpetual software licenses to cloud subscription models, Oracle's traditional software revenue continues to shrink. In theory, Oracle's own cloud software business could capture these migrating customers, but the current market is crowded with numerous players in the cloud software space, and various new AI companies are further expanding customer choices.
Oracle's own cloud software growth is sluggish, with revenue of $4.1 billion this quarter representing only a 10% year-over-year increase, a deceleration of 3 percentage points sequentially. Oracle management indicated that the growth rate for signed-but-not-yet-recognized cloud software revenue is stronger, but investors still have ample reason for concern about the long-term prospects of the software segment.
Industry Briefs
OpenAI CEO Sam Altman informed employees via an internal Slack message on Monday that the company is planning to go public "within the next year." He noted the timing could shift earlier or later, but filing preparatory documents now preserves the option for an earlier listing. Another company executive also revealed that a new major AI model is set to be released soon.
According to Bloomberg, SoftBank's plan to raise $6 billion using its stake in OpenAI as collateral has encountered successive obstacles.
*The Information* reviewed advertiser communication documents indicating that advertisers can place targeted ads for specific products within ChatGPT by simply uploading product information data streams to OpenAI.
Informed sources revealed that the operational performance of Microsoft's Xbox gaming business continues to deteriorate, with plans for layoffs in the coming months. Xbox CEO Asha Sharma stated in an internal email to staff on Wednesday that the division's internally referred-to "measured profitability" had fallen to 3% for the fiscal year ending this month, a trend she described as "unsustainable."
Comments