AI in Marketing

OpenAI has introduced GPT-5.4, its latest foundation model, which is available in standard, Thinking, and Pro versions. The model is designed for professional work, offering enhanced capabilities and efficiency. Notably, the API version supports context windows up to 1 million tokens, the largest from OpenAI to date. GPT-5.4 demonstrates improved token efficiency, solving tasks with fewer tokens than previous models. It has achieved record scores in benchmarks such as OSWorld-Verified and WebArena Verified, and an 83% score on OpenAI’s GDPval test for knowledge work tasks. Additionally, it leads in Mercor’s APEX-Agents benchmark, excelling in creating professional deliverables like slide decks and financial models. OpenAI has reduced hallucinations and factual errors, with the model being 33% less likely to make errors compared to GPT-5.2. The launch includes a new Tool Search system for API tool management, enhancing speed and cost-efficiency. A new safety evaluation tests the model's chain-of-thought, showing reduced likelihood of deception in the Thinking version. Meanwhile, CollectivIQ, a Boston-based company, has developed a tool that queries multiple large language models simultaneously to provide more accurate AI responses. This tool, incubated at Buyers Edge Platform, aims to address issues of hallucinated and biased answers, offering a pay-by-usage model. CollectivIQ plans to seek external funding to expand its market presence, providing a fresh approach in the enterprise AI sector.

Next.

The New York Times reports that Anthropic, an AI start-up, has resumed negotiations with the Pentagon over the use of its AI tools, following a threat from Defense Secretary Pete Hegseth to blacklist the company. The discussions are critical for Anthropic, whose business could suffer significantly if barred from government contracts. The Pentagon, which previously used Anthropic's Claude models in classified operations, also faces challenges if it must remove these tools. The negotiations are complicated by Anthropic CEO Dario Amodei's internal memo, which suggests political motivations behind the Pentagon's actions, claiming the company was targeted for not supporting Trump. OpenAI, a competitor, has added stricter terms to its Pentagon contract, which Amodei criticized as insufficient. The Pentagon insists on using AI tools for lawful purposes, though it may agree to some safeguards. Meanwhile, Anthropic's revenue has surged to $20 billion, but the potential loss of government business poses a threat. In other news, global markets are reacting to U.S.-Israeli attacks on Iran, with energy prices rising and airlines facing disruptions. Additionally, a federal judge has ordered the Trump administration to refund importers for tariffs deemed unlawful, marking a significant development in the ongoing trade policy disputes.

Meanwhile.

AWS has introduced Amazon Connect Health, an AI solution aimed at automating administrative tasks in healthcare settings. This system manages patient verification, scheduling, medical histories, documentation, and coding, integrating seamlessly with Electronic Health Records. It is designed to handle up to 80% of the time-consuming tasks typically managed by staff in large health systems. The AI can book appointments instantly and escalate calls to staff for complex issues. It also reviews medical histories, transcribes conversations, and generates clinical notes and billing codes. Amazon One Medical is already utilizing this solution, which is built on AWS’ Connect platform, enhanced with generative AI capabilities. The solution aims to improve efficiency while maintaining human oversight in healthcare operations.

On a different note.

OpenAI has introduced GPT-5.4, a new AI model that advances the capabilities of autonomous agents by integrating reasoning, coding, and professional tasks such as handling spreadsheets, documents, and presentations. This model is notable for its native computer use capabilities, allowing it to operate a computer and perform tasks across various applications. GPT-5.4 is part of OpenAI's vision for a future where AI-powered agents manage complex tasks online and within software environments. The model can write code to control computers, issue commands based on screenshots, and improve its use of web browsers. It also enhances its ability to call upon tools and APIs more accurately, helping it complete tasks efficiently. OpenAI claims GPT-5.4 is its most factual model, with a 33 percent reduction in false claims compared to its predecessor, GPT-5.2. The model is designed to handle complex queries by persistently searching multiple sources to synthesize clear answers. GPT-5.4 is available in ChatGPT, Codex, and the API, with a specialized GPT-5.4 Thinking model for more complex tasks. This feature is accessible on the ChatGPT web app and Android, with iOS availability forthcoming.

With that said.

A federal judge has denied xAI's request for a preliminary injunction against California's AI training data transparency law, maintaining the enforcement of Assembly Bill 2013. This law requires generative AI developers to disclose summaries of the data used in their models. xAI, founded by Elon Musk, argued that the law violates constitutional rights by forcing the disclosure of trade secrets and being unconstitutionally vague. However, the court found xAI's arguments insufficient to prove a likelihood of success. While xAI continues its legal challenge, it must comply with the law. Other major AI developers, such as OpenAI and Anthropic, have already adhered to the law without issue, weakening xAI's stance that compliance necessitates revealing proprietary information.

Finally.

An artist in their 30s expresses concern over the lack of respect and recognition for their work in the face of advancing AI technologies. Before the pandemic, they had numerous opportunities, but subsequent events disrupted their career and social networks. Their art has evolved to become more narrative and accessible, yet they struggle to connect with audiences without extensive social media engagement, which they find exhausting. The artist is disheartened by the minimal financial returns from their art and the unauthorized use of artists' work to train AI models. They question the value of continuing their artistic pursuits in a culture that seems indifferent to art and artists. Eleanor Gordon-Smith advises the artist to reflect on their initial motivations for creating art, which likely were not driven by financial gain or recognition. She suggests separating the decision to pursue art as a career from the decision to create art for personal fulfillment. Gordon-Smith emphasizes the importance of recognizing the impact of art on individuals, as demonstrated by a fulfilling interaction with a local cashier, and encourages the artist to consider the intrinsic value of their work beyond monetary or cultural metrics.