OpenAI has begun showing advertisements to some ChatGPT users in India, expanding one of its most important consumer markets as the artificial intelligence company seeks new revenue streams ahead of a planned public listing.

The ads are being shown to users on ChatGPT’s free tier and its 399-rupee ($4) monthly Go plan, an OpenAI spokesperson told CNBC. The company said advertising will help support wider access to ChatGPT through free and lower-cost subscriptions while allowing OpenAI to recover part of the substantial cost of running its AI models.

“Advertising would support broader access to ChatGPT through free and lower-cost tiers,” the spokesperson said.

India is among ChatGPT’s largest and most active markets, making it an important test case for OpenAI’s consumer advertising strategy. The company said its advertising feature is already available in 38 countries, including the United States, Japan, South Korea and 31 European countries.

OpenAI said it is taking steps to separate advertising from the answers generated by ChatGPT. Ads will be clearly labelled, visually separated from responses, and displayed below the answer, the company said.

“Answer independence is non-negotiable,” OpenAI said.

The company also said advertisements will not be shown to users who identify as under 18 or whom its systems predict are under 18.

The rollout represents a significant shift in OpenAI’s approach to monetizing ChatGPT. The company has historically relied primarily on subscriptions and enterprise contracts, but the enormous computing cost associated with serving billions of AI queries has increased pressure to generate revenue from its large base of free users.

The economics of consumer AI are increasingly becoming crucial because generating responses requires expensive data-center capacity, including advanced processors, memory, and electricity. Advertising could provide a way to offset some of those inference costs without requiring every user to become a paying subscriber.

Ujjwal Chaudhry, a partner at market intelligence firm Analysys Mason, said advertising is particularly suited to OpenAI because its user base is more heavily weighted toward consumers, while rival Anthropic has a larger concentration of enterprise customers.

“To monetize an individual consumer base on a scale, the ad model is most effective,” Chaudhry said.

His firm estimates that online advertising accounts for more than 75% of global advertising expenditure, with the overall market worth about $1.18 trillion. In-app advertising alone represents a market of roughly $400 billion.

The opportunity is significant, but the move also introduces a new tension for OpenAI. The company must convince users that advertising will not influence how ChatGPT answers questions, especially when AI-generated responses can be perceived as recommendations or authoritative advice.

That is likely to add more weight as OpenAI develops ChatGPT into a broader consumer platform capable of shopping, research, travel planning and other activities where commercial interests could intersect directly with the answers users receive.

OpenAI’s advertising strategy also sets up a sharper contrast with Anthropic, which has positioned Claude as an ad-free alternative. Anthropic previously criticized OpenAI’s decision to introduce advertising, saying Claude would not display ads or sponsored links around conversations and that third-party product placements would not influence its responses.

The competitive difference could become an important selling point as consumers choose between AI assistants. OpenAI has a much larger consumer footprint, while Anthropic has been particularly successful in enterprise AI, creating different monetization strategies for the two companies.

OpenAI is also facing pressure to demonstrate that its enormous valuation can ultimately be supported by sustainable revenue growth. The company is planning a potential public listing in 2027 and has been valued at about $852 billion, according to the report. That valuation places greater scrutiny on OpenAI’s ability to convert its enormous user base and technological lead into predictable cash flow.

Advertising has been touted to become one component of that strategy alongside subscriptions, enterprise products, API usage and partnerships.

Dan Niles, founder and portfolio manager at Niles Investment Management, said OpenAI faces a difficult competitive position between Anthropic in enterprise AI and Google in consumer technology.

“OpenAI is in a tough spot,” Niles said, arguing that Google has a major advantage in consumer distribution as well as the cash flow needed to finance its AI ambitions.

The next stage of OpenAI’s advertising strategy will come on September 4, when the company plans to launch self-serve advertising in India. Unlike the current managed advertising programme, self-serve advertising allows businesses to purchase and bid for ad impressions directly through the platform.

Chaudhry said the model is widely used across social media and could make ChatGPT more attractive to a much broader pool of advertisers by reducing the friction involved in buying campaigns.

Federal Reserve Chair Kevin Warsh delivered his clearest warning yet that the U.S. central bank could raise interest rates if inflation fails to move convincingly toward its 2% target, shifting the policy debate at a time when markets have been expecting monetary easing.

Speaking at the Federal Reserve Bank of Kansas City’s annual economic symposium in Jackson Hole, Wyoming, Warsh said policymakers must be confident that underlying inflation is moving toward the Fed’s objective “clearly and at sufficient speed.”

“Otherwise, we have work to do. That’s our job,” Warsh said.

“The Fed’s predominant focus right now should be on prices.”

The remarks represent a significant change in emphasis from Warsh’s previous public appearances, in which he had avoided giving markets a clear indication of the direction of interest rates. While he stopped short of explicitly calling for an immediate increase, his warning that the Fed has “work to do” if inflation does not move lower effectively opened the door to rate hikes.

Warsh also stressed that his comments should not be interpreted as “forward guidance” and offered no timetable for changing monetary policy.

Still, financial markets interpreted the speech as hawkish. Investors increased bets on a September rate increase following his remarks, turning what had previously been viewed as a relatively low-probability outcome into a more serious possibility.

The shift comes after inflation has remained above the Fed’s 2% objective for more than five years, complicating the central bank’s efforts to balance price stability against economic growth.

Recent inflation data have reinforced the concern. The personal consumption expenditures price index, the Fed’s preferred inflation measure, rose 3.7% year over year in July, while core PCE inflation increased 3.3%. Both remain materially above the central bank’s target.

Warsh’s focus on underlying inflation also suggests that policymakers may be less willing to look through persistent price pressures if they become embedded in the broader economy.

The Fed’s policy rate has been unchanged since December, while longer-term Treasury yields have risen sharply. The 10-year Treasury yield has approached 4.7%, while the 30-year yield has moved above 5.3%, levels that have increased borrowing costs across the economy and put pressure on interest-sensitive areas of financial markets.

Warsh argued that financial conditions may not be restrictive enough to bring inflation back to target.

He said the economy appears resilient and that, given prevailing market interest rates and an unchanged short-term policy rate, “credit and loan markets are showing few signs of policy restraint.”

That observation is important for the rate debate. If financial conditions remain relatively supportive while inflation stays above target, the Fed could conclude that monetary policy is not sufficiently restrictive and needs to be tightened further.

The remarks also come at a difficult moment for the bond market. U.S. government debt has surpassed $40 trillion, while investors have become more concerned about persistent fiscal deficits, rising Treasury issuance and the inflationary consequences of higher energy prices. Long-term yields have risen even as markets have debated the possibility of lower short-term rates.

The Treasury has responded by increasing planned purchases of longer-dated government securities, a move designed to help manage the supply of bonds available to investors. But the strategy has faced skepticism from parts of the market, with some investors questioning whether Treasury buybacks can address the underlying fiscal and inflation pressures driving long-term yields higher.

Higher yields create an additional complication for the Fed. They tighten financial conditions without requiring the central bank to raise its policy rate, but if long-term yields rise because investors are demanding greater compensation for inflation and fiscal risk, the move may also signal that markets doubt the Fed’s ability to restore price stability.

Warsh’s comments therefore place greater emphasis on inflation credibility.

For much of the past year, markets have focused on when the Fed might lower borrowing costs. Warsh’s Jackson Hole speech shifts the question toward whether rates may need to move higher if inflation remains stubborn.

The economic backdrop gives policymakers room to consider such a move. Warsh said the economy remains resilient and that current credit conditions show limited evidence of restrictive monetary policy. That reduces the immediate argument that higher rates would necessarily push an already weak economy into recession.

At the same time, the Fed faces pressure from President Donald Trump, who has repeatedly called for lower interest rates as he seeks to reduce borrowing costs across the U.S. economy and lower the government’s debt-servicing burden.

Warsh did not directly address the political pressure in his remarks. Instead, he framed the issue around the Fed’s statutory responsibility for price stability. His formulation leaves the September meeting deliberately open. Policymakers will have additional inflation, employment and financial-market data to consider before then, and Warsh explicitly avoided committing the central bank to a particular course.

But the message from Jackson Hole was nevertheless consequential.

After months in which investors have debated when the Fed will resume cutting rates, the chair has now made clear that persistent inflation could produce the opposite outcome.

Alphabet’s Google said on Friday that it will change how it applies its site reputation abuse policy in Europe, responding to concerns from EU regulators that the measure could unfairly penalize publishers and expose the company to a potentially significant antitrust fine.

The change follows an investigation by the European Commission under the Digital Markets Act (DMA), the European Union’s flagship legislation for limiting the market power of major technology companies and preventing dominant platforms from using their control over digital infrastructure to disadvantage business users.

At the center of the dispute is Google’s site reputation abuse policy, which targets “parasite SEO”—a practice in which third parties publish content on established websites to exploit the host site’s search-ranking authority and gain greater visibility in Google results.

Google introduced the policy as part of a broader effort to prevent websites from manipulating search rankings. The company has argued that publishing third-party content primarily to exploit a site’s existing authority can reduce the quality and reliability of search results.

EU regulators, however, concluded that the policy could have a wider impact than simply targeting search manipulation. Enforcement actions could demote news organizations and other publishers when their websites hosted content produced by commercial partners, even where the publisher itself had not engaged in deceptive or manipulative activity.

Google said that, from August 30, manual actions taken under the policy to demote websites will no longer apply to users in the 27 EU member states, as well as Iceland, Norway and Liechtenstein. Together, those countries form the European Economic Area.

The company said the policy will remain unchanged outside the EEA.

The decision effectively creates a regional exception to Google’s search-enforcement framework. It also demonstrates how European regulation is increasingly influencing the way global technology companies design and operate products worldwide, even when the immediate legal requirements apply only within Europe.

“We welcome the repeal of this policy, which unfairly penalized publishers and other business users of Google Search,” said Thomas Regnier, a spokesperson for the European Commission.

The Commission said its concerns arose from monitoring Google’s search practices and complaints from publishers. It concluded that the policy could reduce the visibility of legitimate publisher content simply because a website also carried material from third-party commercial partners.

“Thanks to the DMA, Google Search will no longer demote press publications solely for hosting third-party content,” Regnier said.

The Commission said it would continue monitoring how Google implements the revised policy to determine whether the change fully complies with the DMA.

The financial stakes are substantial. Companies found to have breached the DMA can face fines of as much as 10% of their worldwide annual turnover, giving regulators considerable leverage over the largest technology companies. Even without a final penalty, the prospect of such a fine can encourage companies to alter policies before an investigation reaches its conclusion.

For Google, the dispute highlights the difficulty of applying uniform search-quality rules across a digital publishing ecosystem in which publishers rely on commercial partnerships, syndicated material, affiliate content and third-party services to generate revenue.

The same website may host original journalism, sponsored articles, product reviews, job listings, financial information and content created by outside partners.

While removing the threat of manual demotions under the policy in the EEA could reduce uncertainty for news organizations and other businesses that host third-party content, it does not, however, guarantee higher rankings or protect websites from other Google search-quality systems, algorithmic changes or manual penalties based on separate violations.

The change may also alter the incentives facing publishers and commercial partners. Some publishers could become more willing to host external content if they believe the specific site reputation abuse policy will no longer be used against them in Europe. At the same time, Google may respond by developing more targeted methods for identifying low-quality or manipulative content, potentially shifting enforcement from the reputation of an entire website to individual pages, sections or publishing relationships.

The European Commission’s intervention does not mean that Google is abandoning efforts to combat search manipulation. Instead, the company is limiting the reach of one enforcement mechanism in Europe following regulatory scrutiny. Google will still be able to apply other search-quality policies, and the company is likely to continue refining its systems to distinguish legitimate partnerships from arrangements designed primarily to capture search traffic.

The case is another example of how the DMA is forcing major technology companies to adjust products and policies specifically for the European market. Rather than imposing a fine after a full enforcement process, the Commission’s intervention has prompted Google to modify the policy while regulators continue to assess its compliance.

However, the significance of the decision may extend beyond Europe. By requiring Google to suspend the policy in the EEA, EU regulators have created a precedent that authorities in other regions may study when examining the relationship between search platforms and publishers.

Regulators in the United States, the United Kingdom, Australia, Canada and other major digital markets could take similar action if they conclude that search-enforcement policies unfairly penalize publishers, restrict commercial partnerships or give Google excessive control over the distribution of news and online information. They may also view the European case as evidence that search-ranking rules can have competition implications, rather than being merely technical decisions about content quality.

However, other authorities may not adopt the same remedy, but they could use the European intervention as a model for investigating whether platform policies disproportionately affect smaller businesses or media organizations.

There is also a risk that regional differences will make search enforcement more complicated. If Google applies one version of its policy in Europe and another elsewhere, publishers operating internationally may face different expectations depending on where their audiences are located. That could increase compliance costs and encourage other governments to seek their own exemptions or policy changes.