Salesforce is ramping up partnerships with leaders in generative artificial intelligence as investors continue to fear that the software company faces business risks due to the rapid growth of AI.

Just ahead of its annual Dreamforce conference in San Francisco, Salesforce said Tuesday it will enable the use of AI models from OpenAI and Anthropic inside its Agentforce 360 software. A day earlier, Salesforce expanded Agentforce beyond text chats to also handle voice calls.

“The way people are going to interact with software is going to fundamentally shift,” said Brian Landsman, CEO of Salesforce’s AppExchange business and executive vice president of partnerships, in an interview. The interaction could be in ChatGPT or in Slack, he said.

Salesforce will collaborate with Anthropic to bring Agentforce 360 into Claude, Landsman added.

Shares of Salesforce are down about 26% this year, while the S&P 500 index has gained 13%, as Wall Street seeks faster revenue growth from the cloud software company. So far, Agentforce revenue has been “modest,” Morgan Stanley analysts, who have the equivalent of a buy rating on Salesforce, wrote in a Monday note.

Large software companies are increasingly turning to popular AI model developers for new capabilities. Atlassian, Datadog and Intuit have previously signed deals with OpenAI, and Microsoft has invested almost $14 billion in the company. In September, Databricks committed to spending $100 million on OpenAI models.

As part of Salesforce’s announcement, customers will be able to access corporate information in Agentforce 360 and create charts in Tableau through the ChatGPT assistant, which has more than 800 million weekly users. Last week OpenAI announced a software development kit for integrating third-party applications into ChatGPT.

Companies working with both OpenAI and Salesforce will be able to sell products through ChatGPT’s instant checkout feature later in 2025. Salesforce plans to work with Anthropic on selling products for regulated industries, starting with financial services.

OpenAI said last month that ChatGPT users would be able to purchase products from U.S. Etsy sellers and Shopify merchants.

Meanwhile, Salesforce said its engineering organization is adopting Anthropic’s Claude Code programming product.

“We plan to continue to go much deeper with these partners over time,” Landsman said.

Salesforce CEO Marc Benioff has been defending his company’s position in the AI boom. And on last month’s earnings call, he said Anthropic and OpenAI both use Salesforce tools.

“All these next-generation AI companies ranging from OpenAI to Anthropic to everyone are on Slack,” Benioff, who is also Salesforce’s co-founder, told analysts. “And it is incredible how they’ve used that as their operating system and as their platform to run their companies.”

WATCH: Salesforce CEO Marc Benioff goes one-on-one with Jim Cramer

Meta will now limit the content that teenage users can see on Instagram to what they would typically encounter in a movie rated PG-13, the social media company said Tuesday.

With the new content guidelines, Meta said it will hide certain accounts from teenagers, including those that share sexualized content or media related to drugs and alcohol. Additionally, teenagers on Instagram will not be recommended posts that contain swear words, though teen users can still search for it.

The changes come after the company has faced waves of criticism over its handling of child-safety and related mental health concerns on its platform.

As part of the changes, Instagram accounts with names or biographies with links to adult-themed websites like OnlyFans or liquor stores will be hidden from teens, the company said. Teen Instagram users will no longer able to follow those kinds of accounts, and if they already do, they will be unable to see or interact with the more adult-leaning content that they share.

Meta executives said during a media briefing that while the company’s previous content guidelines were already in line or exceeded PG-13 standards, some parents said they were confused about what kinds of content teens could view on Instagram. To provide clarity, Meta decided to more closely standardize its teen-content policies with movie ratings that parents could better understand, the executives said.

“We decided to more closely align our policies with an independent standard that parents are familiar with, so we reviewed our age-appropriate guidelines against PG-13 movie ratings and updated them accordingly,” the company said in a blog post. “While of course there are differences between movies and social media, we made these changes so teens’ experience in the 13+ setting feels closer to the Instagram equivalent of watching a PG-13 movie.”

The social media company has come under fire from lawmakers who claim that it fails to adequately police its platform for child-safety related issues.

The company then known as Facebook came under fire in 2021 when The Wall Street Journal published a report citing internal company research that showed how harmful Instagram was for teenage girls specifically. Other reports have also shown how easily teenagers can use Instagram to find drugs, including through ads run by the company.

Over the past year, Meta has rolled out several features intended to provide parents more transparency about how their teenagers are using the company’s apps. In July, Meta debuted new safety tools intended to make it easier for teenage Instagram users to block and report accounts as well as receive more information about who they interact with on the platform.

In August, the watchdog Tech Transparency Project released a report that alleged Meta’s ties and sponsorship of the National Parent Teacher Association “gives a sheen of expert approval” to its “efforts to keep young users engaged on its platforms.” The National PTA said in a statement that it doesn’t endorse any social media platform, while Meta said at the time that it is “proud to partner with expert organizations to educate parents about our safety tools and protections for teens, as many other tech companies do.”

Meta said its new Instagram content guidelines will begin rolling out Tuesday in the U.S., UK, Australia and Canada before expanding to other regions.

WATCH: Is an AI Bubble Brewing?

California Gov. Gavin Newsom signed a series of bills Monday targeting child online safety as concerns over the risks associated with artificial intelligence and social media use keep mounting.

“We can continue to lead in AI and technology, but we must do it responsibly — protecting our children every step of the way,” he said in a release. “Our children’s safety is not for sale.”

The latest legislation comes as the AI craze ushers in a wave of more complex chatbots capable of deep, intellectual conversation and encouraging behaviors. Across age groups, people are leaning on AI for emotional support, companionship and in some cases, romantic connections.

A recent survey from Fractl Agents found that one in six Americans rely on chatbots and worry that losing access would stunt them emotionally and professionally. More than a fifth of respondents reported having an emotional connection with their chatbot.

Many lawmakers have called for laws requiring Big Tech to better protect against chatbots promoting unsafe behaviors such as suicide and self-harm on their platforms.

The bills signed into law by Newsom on Monday are intended to address some of those concerns.

The changes

One of the laws passed by California implements a series of safeguards geared toward AI chatbots.

SB 243 is the first state law of its kind and requires chatbots to disclose that they are AI and tell minors every three hours to “take a break.” Chatbots makers will also need to implement tools to protect against harmful behaviors and disclose certain instances to a crisis hotline.

The law allows California to maintain its lead in innovation while also holding companies accountable and prioritizing safety, Newsom said in a release.

In a statement to CNBC, OpenAI called the law a “meaningful move forward” for AI safety standards.

“By setting clear guardrails, California is helping shape a more responsible approach to AI development and deployment across the country,” the company said.

Another bill signed by Newsom, AB 56, requires that social media platforms including Instagram and Snapchat to add labels that warn users of the potential mental health risks associated with using those types of apps. AB 621, meanwhile, heighten penalties for companies whose platforms distribute deepfake pornography.

The other key law, known as AB 1043, requires that device makers, like Apple and Google, implement tools to verify user ages in their app stores. Some Big Tech companies have already endorsed the law’s safeguards, including Google and Meta.

Last month, Kareem Ghanem, Google’s senior director of government and affairs and public policy, called AB 1043 one of the “most thoughtful approaches” to keeping children safe online.

Other countries have already enacted rules tougher restrictions on AI. Last year, the European Union passed the AI Act that includes fines for companies that violate the laws’ framework that includes a social scoring systems.

Utah and Texas have also signed laws implementing AI safeguards for minors. The Utah law, for example, requires that Apple and Google to verify user ages and it requires parental permission for those under 18 to use certain apps. These laws have also raised questions over whether harsh restrictions violate free speech or bans are the most effective solution.

California isn’t the first jurisdiction to pass laws like these, but Newsom’s signings carry significance due to the size of the state’s population and the fact that many tech companies are based in the San Francisco Bay Area.

WATCH: Why it’s time to take AI-human relationships seriously