Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• Why Microsoft is using Anthropic over OpenAI

• Senior devs rely more heavily on AI coding

• OpenAI deal with MSFT allows path to IPO

• CA AI bill targets AI companion chatbots

Let’s get right into it!

Sign Up | Advertise | Tools

BIG TECH

Why Microsoft is using Anthropic over OpenAI

Microsoft is expanding its Office 365 Copilot beyond OpenAI by integrating Anthropic’s Claude models, after finding they outperform OpenAI’s GPT models in certain tasks. Internal testing showed Claude produced more accurate results for financial functions in Excel and created PowerPoint presentations that were both cleaner and more visually appealing.

While OpenAI remains Microsoft’s primary partner for frontier AI, Anthropic will power some of the more advanced Copilot features, with Microsoft paying Amazon Web Services—Anthropic’s cloud host—for access. The price of Office Copilot will remain $30 per user per month, and analysts estimate it is already generating more than $1 billion annually.

This decision comes during tense negotiations with OpenAI over its restructuring plans, so the move also reduces Microsoft’s dependence on a single partner and strengthens its bargaining position. At the same time, it addresses customer frustrations with buggy or underwhelming Copilot features by bringing in a model that performs better in practical use cases.

RESEARCH

Senior devs rely more heavily on AI coding

A survey of 791 developers found that senior developers rely more heavily on AI-generated code than juniors, with about a third of seniors saying over half their shipped code comes from AI, compared to just 13% of juniors. Seniors are also more confident in editing and correcting AI output, which helps them achieve greater speed gains, while juniors remain more cautious and often avoid using AI in production.

Still, nearly 30% of all developers say fixing AI code offsets most time savings, echoing findings from other studies showing mixed productivity impacts. Despite this, AI tools improve job satisfaction for 80% of developers, largely by reducing grunt work and boosting enjoyment.

BIG TECH

OpenAI deal with MSFT allows path to IPO

OpenAI announced a new tentative agreement with Microsoft that gives its nonprofit parent an equity stake worth more than $100 billion in the for-profit company it oversees, while retaining control of the business. The deal, framed as a non-binding memorandum of understanding, marks the next phase of their partnership and is intended to help OpenAI raise significant new capital while strengthening the nonprofit’s philanthropic capacity.

Regulators, including California’s attorney general, are scrutinizing the restructuring, citing concerns about governance and the nonprofit’s safety mission. Microsoft, which has invested about $13 billion in OpenAI since 2019, will continue collaborating with the company while both sides finalize definitive contractual terms. OpenAI’s leadership emphasized that the nonprofit will remain in control, ensuring its mission and oversight stay intact even as the organization seeks to scale toward its ambitions around artificial general intelligence.

LEGAL

CA AI bill targets AI companion chatbots

California is considering two major AI bills. SB 243 targets AI companion chatbots, requiring them to avoid harmful content like suicidal or sexual conversations, provide regular reminders to minors that they are speaking with AI, and submit annual transparency reports starting in 2027. It also allows users to sue companies for violations, with damages up to $1,000 per incident. The bill, inspired by a teen suicide linked to ChatGPT, has bipartisan support and would take effect January 1, 2026, if signed.

SB 53, on the other hand, focuses on large AI labs by mandating safety protocol disclosures, offering whistleblower protections, and creating a public cloud resource called CalCompute. It sets tiered reporting standards, with more detail required from labs earning over $500 million annually. While Anthropic supports SB 53, major tech companies like OpenAI, Meta, Google, and Amazon oppose it.

Given the strong child-safety framing and bipartisan backing, SB 243 is more likely to be signed into law, while SB 53 faces tougher odds due to industry pushback and Newsom’s past veto of a similar proposal.