Key Takeaways
- Spammers are turning to generative artificial intelligence (AI) to make their messages more persuasive and evade spam filters.
- AI advancements, such as large language models like ChatGPT, can predict an individual's political leanings, marital status, and life priorities, which could be used to target spam messages more effectively.
- Spam filters can also use AI to create better barriers against unwanted emails by improving their ability to identify and block spam while allowing legitimate messages through.
- In the world of fanfiction, AI has caused a rift in the arts community, with some fans embracing tools like Character.ai bots while others express concerns about text generators being trained on their fanfiction.
- The growing culture war around generative AI has intensified the backlash for supporting or opposing it, and writers must now worry about being accused of using it.
Email inboxes are flooded daily with spam messages from Nigerian princes, wonder drug peddlers, and promoters of can’t-miss investments.
As spam filters improve, spammers develop new techniques to bypass them. Generative artificial intelligence (AI) is emerging as a new weapon for spammers, allowing them to evade filters, grab attention, and persuade people to click, buy, or give up personal information.
Spam is defined as unsolicited commercial emails sent by an unknown entity, which can include text messages, direct messages on social media, and fake product reviews.
Spam is profitable, with one email blast potentially generating thousands of dollars in a matter of hours.
AI advancements, such as generative large language models (LLMs) like ChatGPT, could enable spammers to make their messages more persuasive by targeting individuals based on accessible information like social media posts.
The AI-Enabled Spam Future
Generative LLMs, like ChatGPT, predict tokens (parts of words) in text sequences, and when trained on a large enough dataset, they can perform surprisingly well in other tasks.
Researchers have shown that LLMs can predict which word an individual will say next with remarkable accuracy in a word-generation task called the semantic fluency task.
LLMs can also predict how people will respond to certain types of questions in tests of reasoning abilities.
With these capabilities, AI-armed spammers could make accurate guesses about an individual’s political leanings, marital status, or life priorities.
Furthermore, LLMs could be used to argue persuasively on various topics, making spam even more convincing.
Generative LLMs, like ChatGPT, predict tokens (parts of words) in text sequences, and when trained on a large enough dataset, they can perform surprisingly well in other tasks.
AI Battle in Spam Filters
However, AI advancements don’t just benefit spammers. Spam filters can also utilize AI to create better barriers against unwanted emails.
As AI improves at understanding spam messages, filters could get better at identifying and blocking spam while allowing legitimate messages through.
AI Impact on the Fanfiction Community
In the world of fanfiction, AI has caused a rift in the arts community. Many fans have embraced tools like Character.ai bots, but others have expressed concerns about text generators being trained on their fanfiction.
Recently, a slew of authors on Archive of Our Own (AO3), a popular fanworks website, have received spam comments accusing them of using AI tools to generate their work.
The Organization for Transformative Works (OTW), AO3’s operator, is monitoring the situation and has made changes to the site code to prevent responsibly coded bots from accessing it.
While AI-generated fanworks are not currently prohibited on AO3, the situation is developing, and AI-generated works could potentially violate anti-spam policies.
The growing culture war around generative AI has intensified the backlash for supporting or opposing it. Now, writers must not only worry about competing with AI but also about being accused of using it.
