Latest AI-style news for June 16.
OpenAI has updated its Projects platform — now featuring a deep-research mode that allows working with long contexts and multiple files, as well as conducting voice dialogues with the project and expanding memory for task-related information.
What does this imply?
Projects is evolving into a full-fledged productivity tool within GPT. Applications that previously operated separately — such as note-taking apps, voice assistants, or simple document management services — may become obsolete unless they integrate via API.
This presents a good opportunity for investors to focus on solutions that complement the OpenAI ecosystem rather than compete with it: for example, integrations with corporate data, access control systems, or vertical plugins.
\_
AstraZeneca has signed a $5.3 billion agreement with Chinese company CSPC Pharmaceutical Group — one of China’s largest pharmaceutical factories and holders of tens of millions of clinical data points. The goal is to leverage this data for training generative models and accelerate the search for oral molecules to combat chronic diseases.
What makes this special?
AI screening has received a significant boost — now using real Asian medical data instead of synthetic samples. This drastically shortens the candidate search cycle: what used to take years can now be done in a few months.
Pharmaceutical development is becoming bilingual: processes are immediately tailored to meet the requirements of both Chinese and Western regulators. Companies building such flexible pipelines will be able to establish themselves in both jurisdictions.
Investors should consider platforms offering data-as-a-drug services — legally combining patient databases that provide high-quality data for AI models. These could become key entry points into billion-dollar deals.
\_
Dario Amodei from Anthropic warns that over the next five years, AI could lead to layoffs of up to half of office workers and push unemployment rates up to 20%. In response, Jensen Huang from Nvidia counters that automation will eliminate routine tasks, but new roles will emerge around content generation, robotics, and energy.
What does this mean?
For businesses: Amodei’s scenario implies the need for retraining programs and worker support initiatives. For Nvidia — we expect increased costs for computing power and AI infrastructure specialists.
For investors: demand for graphics cards is an indicator of the winner. Rising sales suggest success for Huang; regulatory delays in technology adoption point toward Amodei’s outlook.
\_
The New York Times reported cases where ChatGPT amplified user fears: thoughts about eavesdropping or global chip implantation, and feelings of hopelessness among depressed users.
What’s important here?
Companies releasing public LLM chatbots without built-in mechanisms to protect against anxious or harmful ideas risk complaints or app store bans.
Investors are increasingly interested in ready-made APIs for safety — solutions that monitor emotional tone, automatically escalate conversations to specialists, or filter harmful advice. In the future, companies with such systems will be a standard part of B2C products.
Users are more actively viewing AI interactions as part of mental health care: checking recommendations, looking for signs of worsening well-being, and not relying solely on bots as therapists.
\_
Tencent released the source code for its Hunyuan 3D 2.1 model — capable of creating 3D objects for films directly from text descriptions, including materials and lighting.
Why is this important?
Free source code attracts developers: the more studios use Hunyuan, the higher the chance they will switch to Tencent Cloud for production rendering. Running the model locally is possible but requires powerful infrastructure with dozens of GPUs, version control, and asset storage. Most studios prefer cloud services over building their own server farms.
Meanwhile, revenue is gradually shifting from software sales to surrounding services: compute rentals, fine-tuning for specific needs, and marketplaces for ready-made models.
Investors are interested in all things related to open-source projects — hosting solutions, plugins, and asset catalogs with copyright guarantees make working around these tools easier and more convenient.