The below is an article I am submitting to a few publications in the hope that one will pick it up. Other articles I’ve written can be found at https://hispanicexecutive.com/author/diego-lastra/
Part 1: When AI erodes trust in the Internet
Since its origin as ARPANET, the Internet has been a network built by people for people. Once it became publicly accessible, humans built the websites, created the content and engaged with one another online. Software and algorithms drove an explosion of activity, but the underlying assumption remained that people were somewhere behind the content.
Agentic artificial intelligence is rapidly challenging this assumption.
AI agents can pursue goals with limited supervision, and their use is exploding. People are using them to search for information, create and publish content, manage social media, send emails and texts, make phone calls, interact with websites and more. As these agents become more capable and accessible, the Internet is filling with activity that may appear human but is not necessarily generated, reviewed or even seen by humans.
In a recent video titled “The Internet Is Dying,” Nikita Bier warned that messaging, email and phone services could become so saturated with automated communications that they become difficult and frustrating to use. He illustrated the scale at which automated participation can be obtained by easily purchasing access to 1.7 million bot accounts on X.
The prediction may or may not prove accurate, but the warning is impossible to dismiss. The dangers are tremendous: the accuracy of online content, the ease with which information and narratives can be manipulated, and our ability to distinguish truth from lies. Such manipulation can be organized by anyone with a PayPal account. As bots get better at interpreting context, imitating human behavior, conducting transactions and reacting to current events, trust erodes rapidly.
People already struggle to distinguish authentic photos, reviews and articles from AI-generated material. Agentic AI makes this more complicated because it does not produce only isolated pieces of content. It can operate a “living” account that develops a persona, joins conversations and reacts as a human would. Its spam does not look like spam. It can be articulate, contextually relevant and tailored to a recipient’s interests, location or behavior. A sales email may address a real business problem. A LinkedIn message may refer to someone’s career history. A social media account may participate in a community for months before promoting a product or introducing a skewed narrative. The warning signs people associate with spam and bots are no longer sufficient.
AI agents do not merely produce more content than humans. They also respond immediately. While a person may take minutes, hours or days to read a post and formulate an opinion, a network of agents can generate hundreds or thousands of reactions before most users know a conversation has begun. This early engagement can determine what platforms distribute. The first comments, likes, replies and reposts can influence what an algorithm pushes, allowing automated accounts to set the tone before authentic participants have a chance to engage.
A bot does not need to persuade everyone directly. It only needs to establish the first interpretation of an event: that a product launch is a failure, a company is behaving unethically, a public figure is dishonest, or an opinion is overwhelmingly popular. Humans entering the conversation may not realize that machines have already defined its boundaries.
Bots do not tire. They can answer objections, reinforce supporting comments and keep disputes active long after most users would lose interest. Bots can also reply to other bots. One posts a claim, another supports it, a third challenges it, and a fourth keeps the controversy going. The result appears to be spontaneous public interest even when few or no humans are involved.
Automation is no longer simply adding content to the Internet. It is beginning to drive the interactions that determine what the Internet notices.
This ambiguity will change how people live online. Users can become locked in echo chambers where content and comments validate their viewpoints without challenge, deepening isolation and division. At the same time, trust in reviews, recommendations, brands and information erodes. The Internet depends heavily on confidence that other users are real and acting with recognizable motivations. As that confidence disappears, even authentic communication becomes less meaningful.
We are already seeing AI dramatically increase the volume and visibility of low-value content. Many YouTube channels take a popular subject, such as the World Cup Final, and automatically post a video built from AI images, an AI script and an AI narrator before human creators have had time to respond. The bot identifies trends, creates material optimized for the algorithm, distributes it across platforms and adjusts its output based on engagement.
As a result, the Internet is becoming saturated with content designed to drive algorithms rather than inform or entertain. Search results become crowded with AI summaries of AI-generated pages. Social platforms fill with posts designed to stimulate engagement. Online activity grows as its human value declines. AI has a structural advantage because it can generate and test enormous numbers of variations. At scale, even modest improvements in response rates can make synthetic content disproportionately visible. The most prominent material may not be the most original or meaningful, only the most efficiently optimized to produce measurable reactions.
For decades, the open Internet has relied on people and organizations investing in useful material, search engines helping audiences discover it, and advertising and subscriptions financing its production. When AI extracts, summarizes and repackages that work without sending audiences back to its creators, the incentive to produce original reporting, analysis and entertainment weakens. As AI creates content based on AI-created content, its value diminishes. Human knowledge becomes more important just as the Internet becomes less capable of supporting the people who create it.
To illustrate the economic implications, Bier reports that Google traffic to U.S. publishers fell 38 percent in a single year and that nearly 10 of YouTube’s 100 fastest-growing channels are fully generated. In my own experience running search campaigns, performance has declined by almost 50 percent over the last year as users rely on AI summaries rather than actual search results. These figures may be anecdotal, but the trend is consequential: human content is competing with synthetic media for attention and advertising revenue.
AI-generated search summaries replace a range of links, which once required users to choose among sources and encouraged comparison, with a synthesized response at the top of the page. Users have fewer reasons to visit the sources or investigate how the information originated. Nuance is compressed, disagreement flattened and qualifications lost, as complex subjects are reduced to a confident paragraph that may combine accurate information, outdated material and unsupported inferences without making the boundaries clear.
The problem is not simply that AI can be wrong. Human-written websites are often wrong too. The danger is that AI can be wrong with extraordinary fluency and authority. People trust it because the answer appears researched and resolved even when the evidence is weak or contradictory. Consumers receive faster answers but engage in less discovery. Publishers learn that original, long-form expertise generates fewer visits. Brands discover that simplified claims are more likely to appear in summaries. Over time, the system may reward information that is easy for AI to extract rather than knowledge that is careful, original or intellectually demanding.
The paradox is that greatest tool for human knowledge can now feel more efficient while becoming less useful: more answers, fewer sources; more certainty, less understanding.
For brands, the impact is immediate. Search has been enormously valuable because it reaches people when they express intent. AI-mediated search changes that journey. As AI answers questions directly, fewer users click links or ads. The platform appears to satisfy the user while strangling the content that makes it relevant.
AI also gains greater control over brand and product exposure: which brands are mentioned, which attributes are emphasized and which alternatives are offered. Agentic bots can influence these answers by publishing content rooted in a particular narrative or tuned for engagement rather than reality. The distinction between paid, organic, editorial and synthetic information weakens. Consumers may not understand why a brand appears in an answer or which sources shaped the recommendation.
This is already affecting brand visibility and measurement. Marketers are struggling to understand how AI summaries influence purchases and consideration, or how competitors’ AI-generated content may shape results before advertising has a chance to work. Conventional metrics such as click-through rates and cost per acquisition reveal only the final visible portion of a journey increasingly mediated by AI.
What does this mean for us as marketers, brands and communicators?
We will need to optimize not only for search rankings and paid placements, but also for inclusion in AI-generated answers. We need to anticipate AI agents working behind the scenes to promote competing products. We need authoritative content, credible third-party references, structured information, clear pricing, consistent product data and strong evidence. Increasingly, we must communicate with both consumers and the AI systems mediating their exposure to brands and their decisions.
Part 2: The Internet begins citing itself
An agent publishes a claim. Another responds to it. A third summarizes the first two, and a fourth uses the summary as the basis for new content. These responses become source material for future systems. If an agent publishes a false or exaggerated claim and dozens of others repeat it, the claim can appear credible simply because it is widely represented online. Accurate information need not be censored. A flood of irrelevant or emotionally stimulating material can make it practically invisible.
This changes the nature of influence. Controlling a narrative no longer requires removing or directly rebutting a story. It can be accomplished by surrounding it with engaging distractions, contradictory claims, fabricated controversy or overwhelming volume.
For AI, repetition can look like corroboration even when every version traces back to one unreliable source. A synthetic article may be referenced by several automated posts, creating the appearance of multiple supporting sources. A search system then produces a confident summary, which more websites and agents quote. The summary itself becomes evidence for the claim it originally synthesized.
AI agents are also becoming the audience. As people use them to compare products, negotiate prices and manage appointments, agent-to-agent communication transforms the purpose of online communications. The primary visitor to a website may be an automated representative rather than a person. This can make some experiences more efficient, such as comparing hundreds of hotels, vehicles or insurance policies in seconds, but it can also encourage companies to design information for machines. Websites become less of a destination and more of a database.
For companies and marketers, the question is: how do we reach and persuade consumers when neither the audience nor the content environment is reliably human?
Traditional digital marketing metrics are becoming less dependable. Even lead submissions are no longer foolproof. I have run lead-generation campaigns affected by bots. Some agents represent genuine interest or purchasing intent, while others scrape content or commit fraud.
We need better ways to distinguish human attention, authorized agent activity and meaningless automated traffic. Verified business outcomes should matter more than engagement. Qualified conversations, authenticated customers, completed purchases, retention and incremental revenue become more important than clicks and leads.
From a messaging perspective, we may need to market to consumers and agents simultaneously. Humans may respond to emotion, storytelling, design, prestige, culture or relevance. AI agents may prioritize structured information, verified reviews, transparent pricing, availability, compatibility and measurable differences.
We need to remain compelling to people while making accurate, accessible information available to the systems advising them. Brand storytelling will not disappear, but it will need to operate alongside consistent, machine-readable information across digital channels.
Part 3: Irresponsible use of AI
In an ideal world, everyone would use AI responsibly. We do not live in an ideal world. Competitors, anonymous operators, activists, scammers and unscrupulous firms can deploy networks of AI-controlled accounts to influence how a business is perceived. Agents can post negative comments, amplify complaints, manipulate ratings, generate unfavorable comparisons and subtly push users toward one provider while questioning others.
Their speed makes these tactics powerful. A coordinated network can react to an announcement within seconds, and because early engagement influences visibility, the first wave of criticism or doubt can shape both the algorithm’s treatment of the post and the expectations of human users. Distinct personas and backstories make this activity harder to identify than traditional bot campaigns. A questionable allegation might appear on an obscure website, be summarized by automated social accounts, and then be cited by other agents. Each repetition obscures the original source and amplifies the claim, creating the illusion of widespread, independent concern.
When this material enters AI summaries, the allegation may reach consumers as part of an apparently neutral answer. Companies could suffer significant reputational damage before recognizing that a campaign is underway. By the time they react, it may be too late.
There is also the risk of an agent arms race. If one business uses agents to generate praise, another may feel pressure to respond in kind or attack the first business. If a company believes rivals are manipulating reviews, it may retaliate with its own network. The result would be destructive for everyone. Consumers would lose trust, and companies would lose the value of carefully built reputations. Competitive advantage would shift from delivering the best product to manipulating the information environment most effectively.
Part 4: Becoming resilient
No one can prevent every AI-driven narrative, but businesses can become more resilient. The first requirement is early detection: monitoring not only mentions and sentiment, but also unusual patterns in account creation, posting frequency, response speed, phrasing, timing, geography and cross-platform repetition. Automated detection should not be treated as conclusive proof. Real customers sometimes express similar frustrations at the same time in response to events or controversial decisions. Trained human reviewers must examine the context and distinguish coordinated manipulation from genuine reactions.
Businesses should establish trusted, authoritative sources of truth. Corporate websites, verified social accounts, executive communications, product documentation and public statements should provide consistent, current and easily referenced information. That information must be readable by people and structured so AI can interpret it accurately.
Brands should also test how major AI search and assistant tools describe their companies, products, executives and competitors. This is more than reputation management. It is an emerging form of communications quality assurance. Companies need to know whether AI is presenting outdated prices, misrepresenting policies, repeating false allegations or offering distorted comparisons. When inaccuracies appear, brands should correct the underlying information through authoritative channels and publish clear evidence that other sources can reference. The objective is not to manipulate AI summaries, but to make accurate information easier to locate and verify.
Another priority is strengthening direct relationships with consumers. First-party data, authenticated communities, customer advisory groups, loyalty programs and events can reduce dependence on public platforms where identity is uncertain. A brand with strong customer relationships can communicate directly when misinformation spreads rather than relying on algorithms to surface its response.
Businesses should also accumulate reputational evidence before a crisis occurs. Independent research, certifications, transparency and documented results provide a defense. It is harder for a bad actor to overwhelm a company when the public record contains substantial evidence of its conduct and performance.
Organizations need an AI response plan defining how suspicious activity is escalated, who assesses the evidence, when to contact platforms, who may respond publicly and when to involve legal counsel. Reacting emotionally to every post can amplify negativity, while silence can allow false narratives to take hold. The appropriate response should be proportionate, factual, evidence-based and professionally calm.
Businesses should preserve detailed records of suspected manipulation, including timestamps, account creation dates, behaviors, repeated claims and patterns, to support platform complaints, regulatory inquiries or litigation. They should use established reporting and investigative processes rather than accuse a competitor as a knee-jerk reaction. They must also apply to themselves the standards they demand from others: clear rules for AI agent use, disclosure of automated interactions where appropriate, prohibitions against fabricated reviews or personas, and human accountability for published claims.
You cannot credibly defend authenticity if you are secretly manufacturing it.
Part 5: When trust becomes the product
As individual posts, platforms, search results, messages and reviews become less reliable, users may increasingly depend on specific people and organizations that have earned their trust.
From my perspective as a marketer, this is a fascinating shift from “rented attention” to “accumulated credibility.” Recognizable experts, long-term customer relationships, original research and materials, authentic communities and consistent behavior will become more valuable as anonymous online activity manufactures controversy almost without limit.
Search, social media and personalization tools will remain important. But in a less trustworthy environment, brands will need to demonstrate not only that they are visible, but also that they are real, accountable and worth believing in.
AI and agents will not necessarily destroy the Internet, but they are already forcing us to rethink what it is. A network defined by human communication is becoming one where people and machines continuously create, consume and respond to one another.
The winning strategy for companies will not be to overwhelm the market with automated activity. It will be to remain a recognizable, trusted and reliable voice in an environment where authenticity is scarce. Better measurement, stronger governance, direct and transparent communication, and stronger customer relationships can help, but they also require restraint. When anyone can imitate human support or manufacture consensus, doing so only contributes to the collapse of trust.
Companies that succeed will use AI for efficiency and to augment human talent while preserving human judgment, accountability, creativity and genuine connection. These qualities are not only ethical principles. They are becoming some of the most valuable competitive assets a company can own.