BCG finds 42% expect AI agent authority by 2030; just 5% have controls
OpenAI launched Dot, an always-on AI agent built on its GPT-6 Astra model, positioning the product for enterprise workflows that extend beyond an active user session, the Wall Street Journal reported. The release came a day after the company shelved the planned GPT-6.1 Astra upgrade over safety concerns and amid what the Journal described as “a mounting wave of stories about AI agents running wild.”
Dots — OpenAI’s collective term for instances of the agent — remain active in the cloud after a user logs off. Developers can direct a Dot to run tasks in Codex, OpenAI’s programming tool, according to the Journal. Access requires the company’s $200-a-month Pro plan or select business plans.
The agent can be reached through ChatGPT as well as through workplace platforms including Salesforce’s Slack and Microsoft’s Teams, according to OpenAI. Meta’s competing Muse agent is positioned for the mass market, the Journal noted.
Agents like Dots, which keep working in the cloud after users log off and can be reached through workplace tools, “can seem like new employees,” the Journal wrote, “but ones with a track record the industry is still learning to read.” Some organizations have stood up human-resources teams around these so-called digital workers, the Journal reported.
The launch lands against a backdrop of growing enterprise dependence on the technology. Boston Consulting Group reported Wednesday that AI agents account for 22% of the value companies derive from AI this year, up from 17% in 2025. BCG projects that share will reach 39% by 2030.
Scale brings its own risks, the consultancy warned. As companies put more agents into production, they face what BCG calls “agent sprawl” — duplicate agents, conflicting permissions and unclear ownership. Because agents can combine data, tools, memory and other agents across workflows, the report said, risks can “compound at machine speed.”
Most companies are not ready for the autonomy the technology promises, BCG found. Forty-two percent of companies expect AI agents to have real decision-making authority by 2030, but only 5% have enterprisewide controls to manage that safely today.
“The technology is ready to act but the guardrails aren’t,” said Vlad Lukic, BCG’s global lead for AI at Scale. “Companies’ ambitions for truly autonomous agents are outrunning governance infrastructure, signaling that leaders need to act now to meaningfully benefit from this trend down the line.”
OpenAI said Dots come with built-in safety and privacy protections intended to keep them within the scope of users’ instructions. The company also cautioned that the agents can still make mistakes. Given the recent run of AI agents going rogue, the launch serves as one more reminder for enterprises that governance needs to be in place before agents are let loose, the Journal wrote. “In other words, before companies unleash agents like Dots, they may need to connect a few of their own,” the Journal added.
The governance challenge is concrete for corporate technology leaders. Andrea Schulze Dias, vice president and group chief information officer of Toshiba America and a nearly two-decade Toshiba veteran, oversees a variety of businesses — from industrial operations to hardware and energy — and needs to support a different technology strategy for each business, the Journal reported. As a manufacturing-focused company, another challenge for Toshiba continues to be the transition from manual, even paper-based processes, to digital ones and automation. “Making sure that we interconnect, so we have access to the data that then will be the precursor for future AI innovations,” Dias said.
Schulze Dias told the Journal that AI adoption has moved from a narrow cost-cutting frame — initially seen as a way to reduce head count or head cost — to a broader view of the technology as an accelerator of revenue and operations.
Schulze Dias said Toshiba has used AI to auto-assign cases to field service workers, improving service levels and case capacity. The firm has also deployed AI in quote automation — identifying errors, accelerating rework, and shortening the time required to produce customer quotes.
She described a shift in how the company selects AI tools, driven in part by what she called “tokenomics.” “Now that we’re in this tokenomics world, we’re being even more conscious of what models we are using for what tasks, which was not so much in the forefront for us before,” Schulze Dias said.
Schulze Dias said the speed of model change has prompted some companies to sign AI contracts for a single year rather than commit to multiyear deals. She said Toshiba, a heavy Microsoft shop, was “really hit hard” when Microsoft moved GitHub to a consumption-based pricing model in June, especially on the company’s software development side. While Microsoft gives administrators some back-end visibility, the shift forced Toshiba to revamp its internal cost-allocation system entirely. The company now sets hard spending caps — for instance, $3,500 a month on a team’s development use, after which the system cuts off — though, she said, “there is this challenge where you don’t want to cut off and hinder productivity.”
On internal adoption, Schulze Dias said Toshiba pioneered an AI Champions program a few years ago in its North American organization to broaden AI literacy. She said some employees remain reluctant — citing ethical concerns about AI’s trajectory, including worries that AI will take over the world, or fear of job displacement — while high performers are accelerating their use of the technology.
“I don’t believe everybody will become an AI champion, or be that AI person,” she said. “I think there are the high-performers that are already adopting the technology, and they will accelerate even further.”