BIG TECH

Apple pushes Siri upgrade further into 2026

When will Siri's promised upgrade arrive? Don't hold your breath, at least not yet.

Since last June, Apple users have been waiting for the promised Siri overhaul, one that would make the assistant more conversational and capable of taking actions on your behalf using personal context. Bloomberg previously reported that Apple had set an internal March release target with the iOS 26.4 update, but a new report reveals the features will now be staggered across multiple future updates.

The delay stems from recent testing snags that revealed software problems, including Siri taking too long to handle requests, according to people familiar with the matter cited in the report. However, some new features may arrive as soon as iOS 26.5, as internal versions of that update already include notices about certain Siri upgrades.

Internal test versions of iOS 26.5 reveal it will include two unannounced features: a new web search feature that functions similarly to Perplexity, and custom image generation. Following Apple’s track record with iOS release schedules, the wait between iOS 26.4 and iOS 26.5 will likely be short.

Yet, the most anticipated feature might not be included. An internal version of 26.5 lets users “preview” Siri’s capability of referencing personal data to add context to prompts, with that “preview” designation likely being a sign that the full-on feature is not ready to ship just yet. 

The report cites a landslide of other challenges, including running behind on advanced commands for voice-controlled in-app actions, early tester accuracy issues and bugs, Siri defaulting to OpenAI’s ChatGPT instead of Apple’s technology, which now includes AI tech from Google Gemini. 

This isn’t stopping Apple’s ambitions as the company is reportedly also working on a revamped, chatbot-like Siri for iOS 27, iPadOS 27, and macOS 27, as we previously reported.

Where Apple is right now in the AI space, timing matters less than execution. After waiting a year and a half, a few more months won't significantly change user expectations. However, Apple does need to work to repair the trust that was broken as a result of the delays and the underwhelming launch of Apple Intelligence. The Gemini integration offers the strongest promise of redemption, signaling that Apple is willing to pivot from its original plan to deliver real value while rebuilding confidence in Siri's promised capabilities. But that narrative only holds if Apple follows through. Getting it right matters more than getting it done quickly. So this slight delay will matter less if the revamped Siri that Apple delivers is finally a game-changer.

TOGETHER WITH PIGMENT

How AI Is Changing Forecasting, Planning, and GTM Execution

AI is transforming how modern Go-To-Market teams forecast, plan, and execute.

On February 26, 2026, join sales and revenue leaders from OpenAI, Spotify, and Pigment for a live conversation on how AI is changing forecasting, planning, and GTM execution in practice.

You’ll learn how high-performing sales teams use AI to improve alignment, make faster decisions, and drive better outcomes. Save your spot and future-proof your GTM strategy.

Save your spot

MARKETS

$30B later, Anthropic doubles down on 'good AI'

Everybody loves Anthropic.

On Thursday, the company announced a $30 billion Series G funding round at a $380 billion post-money valuation. The company said in a blog post that it would use the funding to fuel its infrastructure buildout, product development and frontier model research. 

The round included dozens of investors, with big names such as JPMorganChase, Goldman Sachs, Fidelity and BlackRock on the roster, as well as previously announced investments from Nvidia and Microsoft. The funding is more than triple the $10 billion target the company initially set. 

Additionally, the company announced that its revenue run rate hit $14 billion, a figure that’s grown tenfold annually over the past three years. Anthropic attributed this growth to becoming the “platform of choice for enterprises and developers.” 

In the wake of its success, Anthropic is sharing the love. On Wednesday, the company announced that it intends to cover the rising costs of electricity stemming from the buildout of AI data centers. 

This includes covering 100% of grid updates needed to support data centers, procuring new sources of power to protect consumers from price increases, investing in “curtailment systems” that cut data center power usage, and addressing these data centers’ impacts on communities throughout development. “Done right, AI infrastructure can be a catalyst for the broader energy investment the country needs,” Anthropic said in a blog post. 

The decision comes as Anthropic flirts with building out 10 gigawatts of data center capacity, the Information reported earlier this week. 

The donation isn’t Anthropic’s only show of altruism for the week: On Thursday, the AI firm announced plans to donate $20 million to an AI super PAC focused on safety, guardrails and public education about AI. The group, called Public First Action, has reportedly been in talks with Anthropic about a donation since November, aiming to ensure that OpenAI doesn’t concentrate too much political power, according to The New York Times. 

In a blog post, the company said that the AI policy decisions made over the next few years will “touch nearly every part of public life, from the labor market to online child protection to national security and the balance of power between nations.” 

The PAC’s mission also runs somewhat in opposition to Leading the Future, the political group that OpenAI has thrown its own weight behind, which pushes against state AI regulation in favor of a looser national framework. 

Both the grid investment and supporting the PAC signal that the company is seeking to keep its moral compass aligned with true north.

Anthropic’s M.O. has long been to do AI in the most responsible and ethical way possible (or at least to look like they’re doing so). As a result, the company is building up significant goodwill if or when they do things that might have a negative impact. Still, with tens of billions in funding on the line and its rivalry with OpenAI growing more intense, Anthropic is under a microscope. With more pressure than ever to get returns on its massive investments, upholding its manicured image could become more difficult. 

TOGETHER WITH ASAPP

100 ways to use gen AI in the contact center

Most AI in the contact center talks. Very little of it works.

Real impact starts when AI agents can resolve issues, execute workflows, and operate across enterprise systems.

ASAPP’s 100 use cases for generative AI agents in the contact center shows how leading brands in travel, insurance, banking, retail, and healthcare are deploying AI that delivers outcomes: lower costs, higher CSAT, and measurable revenue lift.

👉 Get the use cases

GOVERNANCE

OpenAI warns against Chinese model copycats

Major AI firms are sounding the alarm on secondhand models. 

On Thursday, OpenAI sent a memo to US lawmakers warning them that Chinese AI firm DeepSeek is using distillation techniques to “free-ride” on the capabilities of OpenAI's models, as well as those of other frontier labs. The firm says DeepSeek is using “obfuscated methods” to undercut OpenAI’s defenses. 

Open AI memo claims that Chinese LLM providers and university research labs are using its models in such a way that would be “highly beneficial” in creating competitor models through distillation. 

• OpenAI also has observed accounts associated with DeepSeek employees using methods to “circumvent” access restrictions.

• Although OpenAI has added safeguards to prevent this distillation, the company claims that these techniques are becoming more sophisticated as a result.

• Although distillation is a commonly used technique in AI training, OpenAI claims that doing this under the radar can result in models that are missing key guardrails, resulting in “dangerous outputs in high-risk domains.” 

“It’s important to note that there are legitimate use cases for distillation … However, we do not allow our outputs to be used to create imitation frontier AI models that replicate our capabilities,” OpenAI said in the memo. 

And OpenAI isn’t alone in calling out these risks. On Thursday, Google’s Threat Intelligence Group published a report detailing a flood of “commercially motivated” actors seeking to clone its flagship model, Gemini. The company said in the report that these actors are using “distillation attacks,” in which they prompt Gemini thousands of times as a means of learning how it works to bolster their own models. 

Though Google didn’t call out any specific group in its report, the company said it “observed and mitigated frequent model extraction attacks from private sector entities all over the world and researchers seeking to clone proprietary logic.”

Despite the risks that Chinese open-source models might present, their cheap and powerful AI models are irresistible. DeepSeek and Alibaba’s Qwen models have raked in hundreds of millions of downloads globally and are even attracting the attention of Silicon Valley AI firms that are using the Chinese tech to build their products. However, if Google and OpenAI are on the mark, these models might not be so different from the proprietary offerings of US firms — just without the essential safety guardrails that set them apart. 

LINKS

• Spotify’s best developers haven’t written code since December

• ServiceNow acquires Pyramid Analytics for $200 million

• AI firm Simile raises $100 million for human behavior prediction

• Ring nixes partnership with Flock Safety after Super Bowl ad backlash

• Confidential AI firm Opaque Systems raises $24 million at $300 million valuation

• Waymo deploys sixth-generation Ojai robotaxis in LA and SF

• Experian in ChatGPT: The Experian Insurance Marketplace app on ChatGPT allows users to access its trusted insurance comparison platform on ChatGPT. 

• Google Gemini: Gemini 3 Deep Think got an upgrade that allows users to urn a sketch into a 3D-printable reality. 

• OpenAI’s GPT-5.3: The company launched a research preview of GPT-5.3-Codex-Spark, a smaller version of GPT-5.3-Codex meant for real-time coding.

• GLM-5: Z.ai released a new model this week that is proficient in agentic tasks, even beating leading models from Anthropic, Google and OpenAI across benchmarks.

• Applied AI Engineer: Production ML systems, LLM integration and enterprise deployment

• Forward Deployed Engineer: Technical implementation specialist, enterprise AI deployments

• ML Research Engineer: PhD in Computer Science, model architecture optimization

• RLHF Specialist: Post-training expert, human feedback integration for frontier models

(sponsored)

GAMES

A QUICK POLL BEFORE YOU GO