Welcome to 9to5Neural. AI is evolving rapidly. We’re here to keep you informed. In our first edition, we delve into the dawn of a new era for OpenAI, Anthropic’s proactive approach to a prevalent AI critique, and the back-and-forth of presidential AI executive orders. Let’s unravel the latest in AI news together.
ChatGPT Teams Up with Operator
OpenAI has unveiled the 18K gold Apple Watch Edition of ChatGPT. ChatGPT Pro, at $200/month, is the subscription service that might make Tim Cook dream of such substantial recurring revenue from each customer.
Beginning today, ChatGPT Pro now offers AI enthusiasts an exciting new reason to subscribe, aside from increased request limits.
Introducing Operator! OpenAI describes it as “a research preview of an agent capable of using its own browser to handle tasks on your behalf.” From creating memes to shopping for groceries and filling out forms, Operator is one of OpenAI’s pioneering agents set to carry out your requests.
Today we’re launching Operator, an agent capable of accessing the web to perform tasks for you. With its own browser, it can examine webpages and engage with them by typing, clicking, and scrolling. This is currently a research preview, which means it has limitations and will evolve based on user input.
Operator won’t always be behind a $200/month paywall. OpenAI intends to later extend access to this AI resource for Plus, Team, and Enterprise users. Presently, Operator can be accessed by all ChatGPT Pro users in the U.S. via operator.chatgpt.com.
According to OpenAI, Operator is driven by its innovative Computer-Using Agent (CUA) technology.
Operator is powered by Computer-Using Agent (CUA), a model that merges the vision capabilities of GPT-4o with sophisticated reasoning through reinforcement learning. CUA is adept at navigating graphical user interfaces (GUIs)—the buttons, menus, and text fields visible on screens—mimicking human interaction. This allows it to perform digital tasks without needing OS- or web-specific APIs.…
While CUA is still in its early stages and has certain limitations, it achieves record-setting benchmark results, posting a 38.1% success rate on OSWorld for full computer tasks, 58.1% on WebArena, and 87% on WebVoyager for web-focused tasks. These outcomes underscore CUA’s capability to operate smoothly across various platforms using a unified action framework.
I suppose now is as good a time as any to announce my departure from 9to5Neural to spend more quality time with my family.
From now on, upcoming editions of 9to5Neural will be powered by Operator. I am confident in the Computer-Using Agent’s ability to convey AI news to the public moving forward.
Wait, hold on. I may have jumped the gun. It turns out there’s a glitch with our ChatGPT Pro subscription. I’ll be sticking around for now!
In all seriousness, Operator is undoubtedly a significant advancement. We may look back on January 2025 as a pivotal moment in AI evolution. The Computer-Using Agent technology could also address the concerns of AI sceptics who keep asking about the arrival of ChatGPT-5.
The other major story coming out of OpenAI this week? Stargate. As Sam Altman noted on X, it’s “big. beautiful. buildings.”
So, what’s Stargate? In essence, it’s a massive computational brain based in Texas. OpenAI provided insights into the initiative this week:
The Stargate Project is a new company planning to invest $500 billion over the next four years in developing new AI infrastructure for OpenAI in the U.S. We’ll start deploying $100 billion right away. This infrastructure aims to cement America’s leadership in AI, create hundreds of thousands of jobs, and generate significant economic benefits worldwide. This project not only supports the re-industrialization of the U.S. but also enhances the strategic capability to safeguard America’s national security and that of its allies.
Initial equity investors in Stargate include SoftBank, OpenAI, Oracle, and MGX. SoftBank and OpenAI are the lead partners, with SoftBank bearing financial responsibilities and OpenAI handling operations. Masayoshi Son will serve as chairman.
Key technology collaborators at this stage are Arm, Microsoft, NVIDIA, Oracle, and OpenAI. Construction is currently underway in Texas, and plans for additional campuses across the country are being evaluated as final agreements are reached.
As part of the Stargate initiative, Oracle, NVIDIA, and OpenAI will work closely to construct and manage this computing infrastructure.
As with any ambitious AI firm, there are backing billionaires, and they are currently in a heated discussion on X regarding Stargate’s finances.
Elon Musk, whose xAI company isn’t connected to Stargate, responded to the announcement on X by stating, “they don’t actually have the money.” Musk further claimed that he knows SoftBank has “well under $10 billion secured.”
Conversely, Altman expresses confidence that the involved parties have their funding lined up.
Additionally, the OpenAI leader noted he fell into the non-playable character trap concerning Trump (now that Trump has made his character playable, in connection to Stargate).
Honestly, I’m more optimistic about the potential of ChatGPT Operator than the complicated interplay among billionaires.
Claude Introduces Citations for References
In the meantime, Anthropic, known for its cautious stance regarding AI safety, is launching a promising new feature called Citations for its Claude chatbot.
Today, we’re debuting Citations, a new API feature that enables Claude to ground its answers in source materials. Claude can now offer precise references to the exact sentences and passages it uses to formulate responses, leading to more verifiable and reliable outputs.…
Previously, developers had to use intricate prompts to instruct Claude to include source information, often resulting in inconsistent results and a significant investment of time in prompt development and testing. With Citations, users can now add source documents directly into the context window, and when querying the model, Claude will automatically cite any claims in its output that originate from those sources.
Our internal evaluations have demonstrated that Claude’s integrated citation functionalities outperform most custom implementations, enhancing recall accuracy by up to 15%.
Anthropic highlights relevant applications, including customer support inquiries and document summarization tasks.
Best perspective? Kyle B. Russel on X, no citations required:
Claude 3.5 Sonnet and Claude 3.5 Haiku are now equipped for Citations effective immediately, and Anthropic has provided documentation for your investigation.
New AI Executive Order Overrides Previous One
Resuming our exploration of presidential politics, let’s take a look back at U.S. AI policy.
President Trump continued his series of executive orders on Thursday by revoking the Biden administration’s AI policy in favor of his own.
In case you’ve forgotten, Biden’s executive order focused on the safety of AI technologies, establishing infrastructure standards, addressing job displacement, and enforcing watermarking for transparency of AI-generated content. In summary, Biden’s executive order:
- Prioritized the safe, secure, and trustworthy advance of artificial intelligence (AI).
- Required standards for crucial infrastructure, enhanced cybersecurity, and oversight of federally funded initiatives.
- Addressed societal issues, including job disruption mitigation, equity advancement, and civil rights protection.
- Mandated that AI-generated content include watermarks for clarity and differentiation from human-generated material.
According to an AP report, Trump’s new AI executive order nullifies prior government regulations viewed as obstacles to American innovation in AI, insisting that the U.S. “develop AI systems that are free from ideological bias or engineered social agendas,” as spelled out in the order.
Additionally, the AI executive order authorizes the creation of an AI action plan within 180 days, led by Special Advisor for AI and Crypto David Sacks, a former PayPal executive appointed by Trump.
Going forward, tech companies will no longer need to disclose their development of AI models that surpass a certain power threshold to the government.
Intense Competition from DeepSeek R1
Meanwhile, the race for AI superiority extends beyond U.S. enterprises. This week, the Chinese AI firm DeepSeek launched its R1 model family.
What’s remarkable about R1 is its ability to run locally while matching the performance of OpenAI’s ChatGPT 4o model. Typically, local models lag behind cloud-based offerings, making this development from DeepSeek noteworthy.
The drawback? R1 evidently adheres to an approved state viewpoint regarding historical events, particularly concerning subjects like the Tiananmen Square protests of 1989 and Taiwan’s sovereignty. It’s a clear reminder of the stakes involved in the AI race.
Stay tuned for more insights on the latest AI advancements in the next edition of 9to5Neural—exclusively on DMN!
Top iPhone Accessories
: . More.
